Techstrong TV May 19, 2025

Hey everyone, it's Tim Cook in Hot Water with Duche. You're watching Textron Gang. Hey everyone, happy Monday.

I hope you had a great weekend. It was a great weekend. I was up in Boston to actually, my son, my oldest son, graduated law school and it was quite a, uh, well, it was a proud moment for me, so all good there.

Um, but we're back here on the gang and we've got some great gang members. Let me get right to them. First of all, joining us from some airport somewhere in the world.

He is a future analyst, visible impact principle analyst, Techron loyal Techron gang member, moving up the ranks here, the one and only Guy Currier. Hey Guy. How are you?

Hi. Good to be here. Yep, I'm good.

I'm in the Orlando airport, the wonderful Orlando airport on the way back. 'cause I spent the weekend, uh, with my uncle who lives in here. And, and, and also I do wanna say my daughter is starting college, uh, next year, but, uh, her plans to go to Ross School, so congratulations to You.

Thank you. You know, I mean, he did the hard work. I just wrote the checks, but, um, nevertheless, very proud.

It's a proud moment. All right, moving over from guy, one of our newest members of the gang, but she's been on three or four already. She's a regular, excuse me, cybersecurity expert extraordinaire, Teri Robinson.

Hey, Terry. Good to see you. It's so good to see you.

And I wanna extend my congratulations, uh, too. Congratulations for, uh, son's graduation. Uh, my daughter graduates, uh, today, Monday afternoon, uh, from law school as well.

So, um, Really good for you. Nick stopped the bar, I guess, right? Yeah, yeah.

I think well in bar. So he's taking the mass bar up in Boston, and it's the end of July. Yeah, they're all around The same.

That's when she's taken the New York bar, so, Yeah. Very cool. Very cool indeed.

All right, moving over to Silicon Valley where he looks, he looks downright Roy in the cheeks today. Even he, I don't know, maybe, I don't know if it's makeup or someone's been pinching them, but he's our Silicon Valley editor, Jon Swartz. Hey, John.

How are you, man? I'm good. I'm back at the friendly confines of Belmont.

Congratulations to all our college graduates and people going into college. Um, uh, it's great to be here. Absolutely.

It's great to have you on. And then finally, he might be graduating from going to minor league games to the big league. Well, Mike, you, you had the subway series this weekend, but, um, we're here not to talk baseball, but Juan Soto got booed really bad in it, and I gotta tell you the truth, I felt good.

Um, our Chief Content Officer, Mike Vizard. All right. We got, we got two outta college, one in and one to go.

So I'll be working for a while. Yeah, That's what that means. Um, anyway, let, let's move into it.

I, I kicked it off. It seems our friends at Apple have, uh, have crossed the MAGA line. You know, um, they, well, they're not gonna, you know, they're not gonna build more iPhones in China, but they wanna build them in India instead.

And, and, you know, uh, while he was in the UAE giving them 500,000 GPUs from Nvidia to build the biggest a AI center there, he, he took the time out to chastise Apple and Tim Cook about building iPhones in India. Mike, what do you think? Well, it seems like, um, to your point, we talked about the whole, uh, meddling in the GPU business last week, and now we're extending that out to smartphones.

And he apparently told, uh, the CEO of Apple that, um, manufacturing in India was undermining the trade position for the United States government. So all these things now are somehow other bargaining chips. But guy, this is unprecedented in my mind, but what's your take?

Well, I think what's unprecedented is the degree to which just lip flapping is being taken seriously because of the position of the person flapping lips. I mean, I, I I, you know, last week, um, the Trump was out in, uh, in, in the Middle East being, you've heard the phrase that being treated like royalty, but in this case take it literally, literally being treated as if he were a prince, um, or a king or what have you. Um, and, uh, I, I completely believe that he had the, the conversation with Tim Koch.

I do not necessarily believe liter. I don't definitely don't believe literally what he said about that conversation. Um, uh, other than the detail of saying, Hey, you know, you, uh, you have to manufacture everything here.

Um, I think he probably did say that. Um, and, uh, apple, I mean, pity the poor tech billionaire, CEO, right? Um, who has founded, uh, in Tim Cook's case, a lot of that billionaire ship and the, the deserved success of Apple on manufacturing and assembly of, uh, iPhones, but also a lot of other devices in China, and is trying to find some way to find stability in supply chain, um, in the midst of complete instability in, uh, trade policy and, uh, to actually, tariffs is the usual word, but all kinds of uncertainty over how much, uh, costs of manufacturing, assembly overseas are gonna be, right?

So they move to a different wait, what would anybody do in that position, billionaire or not? They're gonna move to another source. And they've increased the investment in India from, I mean, John, you wrote this story.

I think they're moving towards something like 25% of iPhones being manufactured in India, 2015. Yeah, 25%. Yeah.

Is that 15% a few years ago? Um, but let's call this for what it is, which is, uh, another thing for, uh, a politician to say and take credit for that doesn't necessarily have any actual foundation behind it. Our president happens to do that about 99% of the time, instead of 50% of the time, or 20% of the time.

You know, how civically you feel. I would not take this to the bank in any way, except for one thing. We still don't really know.

We don't dunno what Tim Cook said. We don't know what Apple's gonna do. All we have are the words of somebody who says all kinds of things all the time.

And, and I, John, I really appreciated the way your story was liberal with the quotes of our president. Um, instead of trying to take them at face value, just present them to everybody. Yeah, no, it's like a mob boss, you know, trying to strong arm somebody and shake 'em down.

I mean, it's just the, this is the think of, I think of the absurdity, and I know Alan's gonna weigh in on this. We've got maybe arguably the greatest CEO in terms of operations and running a company who's trying to do the best thing for his shareholders and, and his employees, employee and customers. I'm Employee and customers too.

He's trying to diversify the, and customers, especially customers. He's trying to diversify the product lines so he doesn't jack up the price of the product. Um, he's keeping his options open.

And in fact, from what I understand, within Apple, they have, um, plans in place to either move harder to India or have to, or fall back into China. What, what is must, must be so galling to cook is this idea that Trump is gonna probably change his mind yet again. I mean, he dropped the tariffs against China.

He doesn't change his mind. John, John, he doesn't change his mind. These people take a stick and hit him on the head and say, stupid back down.

He pulls up like a cheap suit. I'm, but here's my, here's my Change his mind. But here, for, for, for Apple, the fear, the fear is they have this exemption in place, right?

He's gonna lift that probably, I mean, he's gonna change his mind. Yet again, it's a guy who, who, who doesn't understand basic math trying, telling somebody who's a genius at operations. I'm just gonna play games with you, despite your commitment to spend hundreds of billions of dollars on the US over the, over four years in manufacturing, what have you.

It, I mean, the whole thing is so galling and it must be chilling to any other tech company or, or any company that has to deal with this, this BS and it Can't going out. Well, I do think its a shakedown. I think your essential point is it's a shakedown.

And the, and, and the, the, the important thing is that the shakedowns never stop. They never stop. So if you give in one, no, Here's the, I doesn't mean they'll stop the important, it's not just a shakedown.

It's not just, it's a shakedown look, should Tim turn around and buy him a nice plane, and then everything will be okay, right? Is that what it takes? Because, and you know what, for with all due respect, Camilla Harris brought this up in the debate with him.

The one debate, 'cause he was afraid to do more 'cause she ran around him. That foreign leaders know his game, they know flatter him or flat out bribe him, and he lays down like a puppy dog to scratch my belly. Right?

That's what you're dealing with here. But let me, let me sit down, let me unpack this for you. Besides buying him a plane or something that'll mollify this guy and let you do anything you want.

Let, let's look at what it is. This, this. SOB is in the Middle East cavorting with people who have financially supported terrorists, who have blown up stuff and killed people in the us, let alone in Israel and everywhere else, right?

It's a known fact that the Qatari support Hamas, it's a known fact where Bin Laden's financial support came from, right? But is okay. And he's going to use Apple as a whipping boy, a fine corporation that's already pledged $500 billion.

He's going to use them as a whipping boy while he allows UAE to, to, as I mentioned, get 500,000 Nvidia GPUs to build an AI campus. They're perhaps the biggest in the world while him and his sons are making $400 million deals to buy golf, to build golf courses and build buildings over there, right? Didn't think it was important enough to go up to Turkey and deal with anything around this Ukraine, Russian war.

'cause that's really just lip service until Vlad pulls his string and tells him, get over here. Right? This whole thing is ridiculous.

But let me give you another thing. Apple's duty is not to the United States of America, though they're a US company or a US based company. They have customers and interest around the world.

Tim Cook has one boss, and it ain't Donald Trump. It's the board of directors of Apple and their shareholders. And he has a fiduciary duty to do what's best for Apple and their shareholders, not even their customers, their shareholders.

And that may or may not be in sync with what, what's best for us policy. But this is what happens when you have global corporations. There's a, and it's high time.

We remember corporations have a duty to their shareholders. If Tim Cook starts doing what's right for the US at the, at the cost of his shareholders, he's going to get fired and sued and, and right for himself. Well, Imagine, imagine if, if, imagine if Trump had his way.

So, uh, say Apple and, and, and in, in just in she panic was forced into doing this. It's, it's, it's absurd. It would take years, first of all, to ramp up production.

It's no, it would cost 3,500. It's for an, but it's like, it's, but here's the visual. You got this 78-year-old felt, 280 pound person with Jensen Wong, Elon Musk, and all these other syco fonts glued onto his hip or his backside or whatever, and he's telling him, yeah, give him 500,000 GPUs.

Give them some starlink. I want my name on that building. Give me this plane.

Dude. The, did you ever watch the Star Trek episode where they land on the planet, that somehow there was a book like Al Capone Chicago was left there, and 500 years later, the whole civilization is built around the gangs of, of Chicago. This is what we live in.

This is what we live in. It's, it's, it's ludicrous. It's psychotic.

Stop, stop the nonsense. Hey, I Have an idea. Maybe, maybe Apple.

Maybe Apple, maybe Apple should build a special Trump iPhone like they did for you two. Maybe they, they, they appeal to his ego and narcissism and they build a, a trump A key on this. No.

I'll give you a better example. Do you remember Godfather part two, Michael Corleone and Hyman Roth go to visit the, the president of Cuba, uh, Duarte, right? I think, no, that, that was a different Banana Republic, whoever was the president before Fidel took Batista and Batista, they're sitting around these table with all these captains of industry, much like Trump is.

And he passes around a solid gold telephone as a gift to Batista from, from ITT. And everyone look around and everyone looks at the phone and they pass it around. And it's very, it's the, this is the same.

It, it's like out of a frigging movie, dude. Come on. This is, we're living, we're living in a, in a nightmare.

We live in a nightmare of what the US should and could be. That's the bottom line. You know?

You know what was really also telling Alan is when you mentioned the sink offense, we had the, the president of Alphabet of his I-B-M-C-E-O, Sam Altman was their Qualcomm, CEO. Well, I mean, it Was a little disturb. Well, you don't think it's the same thing when Putin comes somewhere and he brings gas prom and that prom and the other prom, and they all go to a prom.

It's the same, it's the same thing. It's the same thing. Um, I just, I, you know, I'm, I'm not, I'm not, I'm not really sure though.

You're characterizing the nightmare correcting. I think there's a lot nightmarish about this. But in it, in the context of this show, this podcast, the nightmare is that there are businesses and individuals around the world who rely on Apple and the iPhone in particular to conduct business or to live their lives.

And they have no idea. It's not that they think the price is gonna go up or down or whatever. They, they, that whole platform for them has become less reliable in terms of cost use, what have you.

And that may be one piece of it. When you think about the fact that these kind of, this flapping of the lips can happen anywhere with any company or country randomly, and continue to create that uncertainty. We don't even have to talk about business uncertainty.

We can just talk about the uncertainty of individuals around the world being able to do the things that they wanna do. That I think with technology in the, in the context of this show, so the rest of the nightmare, you're talking about having a kleptocratic or godfather as a president, or whatever it is. I agree with you, that's a nightmare.

That's a political nightmare. It's a global economic and global political nightmare. But I, I always try to bring it back to the fact that this is text draw.

And I would like people to understand like, what do I do now? Great. The flat slips, what do I do?

And I think it's just recognize that the problem you're facing, you, well, I buy, you know, what to begin with. Recognize what the problem is within this context of technology planning is gonna be more. Well, I just, I was just gonna say, I don't, you know, I don't think this is good for anybody that trumped himself, and that's only a temporary flash, I guess.

Um, he roiling the markets worldwide. That's what he is gonna do. That uncertainty isn't good for the economy.

It's not good for the tech companies. It's not good for the country. And it's certainly not good for who you're talking about guy, the people who have to use this stuff.

So, um, I just wonder at point, this breaks, what, when does this break? When does it break? When is it enough?

Or when does something happen? There's like a, a critical mass and, and, and it just, it falls apart. I don't, I mean, am I being hopeful to think that that might happen at, at some point?

Um, and, and there's a, a shakeup in our government or somebody. I don't, I don't even know how you leash him or pull him in from this kind of stuff, but he's, there's just gonna be such long term, uh, impact on, on everything. I don't mean to sound.

So my next question is, is, is there gonna be a shakedown of Samsung to follow this? Because that seems to be the next thing. I mean, have Own, well, he doesn't consider Samsung an American company, right?

The concept of an American company versus those guys. We made t we made tm SC commit to building plans here in the US suspect. Well, because we hold the gun to their head that we won't help them when the, the, the Red China people invade, right?

I gotta bring this back To the, the, I gotta bring it back to the flapping of the lips. Idea, purpose of the statements is publicity trump's own brand image, power. It's not an actual move of any kind, other than that.

So, no, it's not Samsung, because, you know, I I don't think it's Samsung either, because to, to Samsung is not this burnish brand, this amazing brand like Apple. So name a bunch of amazing brands, and yes, but Samsung. Samsung, However great Samsung is not Samsung is not an American company.

But there's one other aspect that we haven't touched on. If Donald Trump was in the White House press room or here domestically, and this is a domestic thing, it would still get a lot of play. But, you know, he's here doing his thing.

But no, he was actually at a press conference with the, the, the Amir of UAE and he brought this whole thing up. The Amir's looking at him like, what, what? Have you seen the video of this, the visuals?

They're looking at him like, where, because he also at the same press conference brought up, what a great job he is done with the price of groceries. And he said to the Amira, I don't know if you know what groceries is, it's an old word or something like that. This this buffoon right?

Is out here in the world not realizing the jokes on him. They're not laughing with him. They're laughing at him, but they know all they gotta do is throw a couple dollars his way.

And like I said, he rolls over. And, you know, one thing that the, the oil rich Arab states are really good at, it's why they're still in power. These Amirs and sheiks and all of them, they've always bought their way in.

They've always bought their way in. Have we forgotten what happened to the reporter? Aan k Kgi kgi, right?

Yeah. You don't hear about that anymore. He's bought his way in.

They've done this. This is what the Amirs and the Sheik and the, the, the Saudis do when they find someone who's viable, they buy 'em. And that's what, and, and be proud America, you've been, your president's been bought.

Let's take a break. We'll come back, we'll talk about csa. Hey folks, we're backing.

Well, the fun never ends when it comes to anything related to Washington. And there's been another turn in the C of drama, it's a bit of a soap opera these days, but apparently there was a contract that got canceled that was, um, being disputed anyway, but now we're just basically saying, well, it's all mute. Because anybody who was involved in actually needing that service from a company called Lidos, well, they're not working there anymore.

So, Terry, what is going on in CSA from your perspective? And is this gonna be just, you know, continuing drama? Sure.

Uh, well, you know, more shenanigans basically from this administration when it comes to, to CS a, I mean, I, I think really all of this stems back right to, um, uh, when Chris Krebs said that the 2020 election, uh, was secure and, um, that sort of undercut, you know, Trump's narrative that, uh, he had, uh, really won the election and whatever, um, and Joe Biden had lost. So I think, you know, going back to that, we know that he's going after Krebs already for, um, you know, what, whatever it is. Um, I, I don't think there's any case there, but, um, he's publicly trying to skewer him.

So I think this is a bit more of the same. 4 billion contract was under dispute, um, already in the courts. But that's because Knight Wing, uh, who was competitive for this bid, um, uh, and, and, and Lados, um, were sort of duking it in.

But, um, the Department of Homeland Security has made it clear, or they've said that this has nothing to do with that. It's, uh, they're pulling the contract simply because the nature of CSA has changed, the personnel, the mission and, and whatever. Um, and it's a, it's a shame.

I mean, this, this contract is, uh, for agile cybersecurity technology, uh, technical solutions, right? So that supports, uh, analytics, testing, integration of security tools. Um, at, at this point, it's, it's gone away.

Uh, honestly, it feels to me, um, again, like it's, it's just this, uh, administration's grudge against CS a and also, uh, kind of fits into what Chris Murphy, uh, I know you guys saw that how he secured Christie, uh, no last week. Um, and said that basically DHS has turned into sort of the, the border security. I mean, that's where all their money is going.

That's where all their effort is going. I think they don't care about taking, uh, uh, money away from cybersecurity efforts at a time when we probably under heightened threat. So, look, you put a puppy killer in charge of these things, right?

And this is what you're dealing with, again, the, the, the ludicrous nature of our lives these days in the us right? You have a woman who's vastly, vastly underqualified here. Other than that, she sucks up to Donald Trump, and she shot a puppy once.

She didn't like it. You put her in charge of the DHS, and, and you make their primary mission to be border control, as you mentioned, Terry. But it's not just the DH S'S primary mission.

We're taking away troops in Europe and all over the world, because the primary mission of our armed forces are also to be border control. We, we have this is, this is like outta the, you know, I feel like an iron curtain has descended upon the continent, right? That that's what you, this is where we are now.

Csa, which has probably done more to help cybersecurity than any government agency or initiative that I've seen in my 30 years of doing this, has been bludgeoned bludgeoned. I mean, she, she, but here's the thing. You gotta give these people credit.

They're so transparent, they don't even hide it. She got up at RSA and said she was doing this and, and proud of it and proud of it. People are sitting in the, the audience.

I was there. They're sitting there shocked that this woman has the audacity to come to RSA conference where the security world gathers to tell us that they're, they're disarming csa. So what do you expect?

What, what, you know, when someone has said this before, if it walks like a duck and quacks like a duck, it's a duck. If it sounds like a fascist and acts like a fascist, they're fascist. That's what you're dealing with.

So, I I, I surprised by this. No, I do hope that private industry can fill the vacuum private things like R-S-A-C-R-S-A community and some other Linux foundations and, and some of these non-governmental entities, because there is a vacuum to be filled here. 'cause I fear we are going to have some sort of critical infrastructure or some sort of catastrophic cyber warfare incidents because we, we are taking down, we're, we're laying down our arms, we are laying down our arms and saying, as long as you're not brown and coming over the southern border, we don't care what you do.

The part of this that's crazy though, the, the, the part of this that's exceptionally crazy is that there's sort of this assumption coming from them that says, um, you know, those attacks from Russia or China that involve any kinda, uh, social engineering and propaganda are not attacks. Right? Those are not within the purview of CS a to go defend.

And that part is crazy because you're allowing foreign entities to pollute conversations and attack election systems with nobody there to defend it. Yeah. But if you, You know, it's almost like It's open, open now.

We're wide open. Everything's Backward, everything's backward. We, we look the other way, yet we accuse CSO of being Ministry of Truth, and we denigrate Chris k Cribs, who was great at his job while we, we dismantled this organization.

Sorry, sorry, guy for interrupting. No, no, no. I would say it struck me.

Um, the, the, uh, which I think Al alluded to this really straightforward nature of this. They didn't say flight. They didn't go quite there and say, we don't need this $2 billion contract because we, the agency that would use it is basically destroyed.

And, uh, it's, and stood down. But they said the priorities have changed at cisa, which is Right. That's actually right.

They didn't even say, oh, we're gonna save a bunch of money. And one, one thing that was really interesting, uh, Terry, uh, you mentioned the, the, the Chris Murphy, uh, and the, just the general engagement that, um, that Christie Nome had, uh, at, um, uh, with Congress last week or two weeks ago, whenever it was, um, because he pointed out to her that they've spent all their money for the year, more or less already. Yeah.

They're, they're gonna be spending more than they have, and they're not to do that. So They'll spend more and then, and then they'll talk about what a terrible deficit the previous administration left them in. Yeah.

But that's a whole other thing that's going on here. They're not proposing budgets for next year. They're spending in, you know, way like, what was it about a $500 million ad campaign that, uh, that, that they put up at, uh, at Homeland Security?

I don't know how much of their budget that is. Probably not a whole lot, honestly. And canceling a a multi-year contract might sound really great with all these billions of dollars and stuff.

That's really not gonna make the whole, it's all theater. It's all theater. I'm gonna, I'm gonna start talking like Alan now.

It's all theater. The guy they put in, in charge of fema, which I believe also is Homeland Security. He, he came out the other day and said, look, he's never done this before.

He really doesn't have a hurricane or emergency response plans put in place. They've lost a lot of people. Give him some time.

He's working on it. It's hurricane season here in two weeks. Well, yeah.

And as What, what catastrophe, what could go wrong? Um, I'm sorry. Go ahead, Terry.

I was gonna say, as a native of Louisiana, that actually makes me shutter. Um, but, but I think, you know, to your point though, Mike, too, I mean, if you, if you admit that these things are going on that Russia's doing this, or China's doing that, or you know, north Korean hackers or whatever, then you are admitting that they are our enemies and that they probably had something to do with, you know, election interference previously and all that. This administration can't admit to, to any of that.

So I think CISA is gonna, whatever is left the cease is gonna always be in the crosshairs. And I think we're not gonna, uh, you know, really admit publicly to these problems until we have some big incident. And to your point, you know, Alan, yes.

It's like an iron curtain dropping down, but what can penetrate that iron curtain? It's cyber attacks. Yes.

Yeah. Crazy. All right, let's take a break.

Can we talk about something not related to this administration? My blood pressure is, is high. You are watching Textron Gang.

Let's come back and talk about Salesforce or something. Discover Textron Group, the epicenter of tech innovation. We are your go-to for reaching IT, leaders and practitioners worldwide.

Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients.

Let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Techron Group. Hey folks, we're back.

And we're talking about an acquisition that Salesforce made where they bought a company that specializes in AI agents. It kind of took me by surprise. 'cause I could have sworn that I had just spent the last two years listening to Salesforce about how great they are and AI agents, and then they turned around and bought somebody who, well, looks like they're pretty good in AI agents.

But John, am I being cynical? Uh, no, you're not being cynical. You're being brutally honest, Mike.

Um, thi this seems to be a pattern, right? Where we're seeing the, all these AI related acquisitions to kind of fill gaps or add spackle to what companies have told us they can already do. So it kind of raises every, is a little bit of suspicion.

Um, so Salesforce basically says they've, they've agreed to acquire convergence, um, dot ai, which is behind these advanced systems that perform complex human-like tasks in digital environments, such as AI agent design, autonomous task execution, adapted systems. And in a sense, basically they're, they, they acquisition is meant to advance what agent force, the foundation of Salesforce's AI strategy is supposed to be doing, or has been doing, maybe struggling to do. It's interesting, I think Mike, a couple weeks ago you wrote a story.

Um, Salesforce has been just clobbering us over their head with announcements, but they, they, uh, they made this, uh, they had this continuing EGI initiative that they've been pushing, and, um, they, uh, made us a couple of announcements, including CRM Arena. And, and I, it's, to me, it's interesting. I I think we're gonna see more of this.

I mean, literally every other day we're seeing a minor acquisition, uh, by a large company that claimed to have a comprehensive AI agent strategy. That's my takeaway. Um, so I too share your cynicism, right?

It does feel like a rollup is already starting to happen in all these AI startups where the bigger companies are just buying them. I guess maybe they figure they're less expensive now than later, or maybe in the pricing is right. But, um, I don't know.

Alan, do you think the incumbents have an advantage when it comes to agent AI because they have all the data, or, you know, is this gonna be, the startups are eventually gonna take over 'cause they're just faster and better innovative? So, you know, last week, last Thursday I did one of my shimmy says my weekly shimmy says, uh, on LinkedIn, there might be a short of it on YouTube. I'm gonna get these all moved over to text on tv.

I address this, right? I addressed this issue, spoke about this one a little bit. Um, Mike, the startups never really take over.

If you are lucky with every wave, one or two of the startups make their way into the old boys club, but it's still an old boys club, right? And this is the way of it, right? These startups innovate the little fish, they innovate the medium fish.

They, they buy that innovation and they make products out of it and product market fit. And then the big fish eat the medium fish, and they put those products into their platform. That's what this is, right?

And as a Yankee fan, Mike, I'm surprised that you don't see why Salesforce would buy conversion. You could never have enough pitching or left-handed hitters in Yankee Stadium. And you, if you're Salesforce, right?

If you're Salesforce, you never have enough ai uh, agentic AI expertise. But here's The thing and why Yeah, go ahead guy. Go ahead.

No, no, You go, Come back. No. Why do you need that stable?

Why do you need that stable? Because, you know, the number of baseball scenarios is infinite. And so you're just trying to cover as many as you can with different, even, you know, uh, uh, two lefthanded slider pitchers can have different approaches that you might wanna use.

One may it hard. So you have an infinite bullpen. You would, I can, I just wanna, I just wanna help us step back a little bit on this.

I spent, uh, uh, the last two weeks at two successive conferences that talked about AG agentic ai. And do you remember on this show, I've said before, there's no such thing as an AI application. I posted for rants about this before.

There's applications and they use ai. So Agentic AI is an agent that uses AI and one or more AI services, but it's an agent, no CIO or CTO, everyone around like five years ago saying, Hey, you know what we need? We need more agents.

The question is, the question was for what? For what? I need an ops agent.

I need, you know, maybe a, a a workflow agent. Like there are lots of agents all up and down the stack. So to say ag agentic ai, it's just a real non-descriptive category.

Why did Salesforce buy an a, a, a agentic AI specialist? This is this human thing, right? This, this, they act as humans, right?

That gives me the willies, but let's just go with it. Um, well, I maybe ultimately the idea is to replace entire Salesforce with Salesforce. I don't know.

Well, how about they did it? Because they can. Yeah.

So, so there's just, yeah, I'm gonna disagree in the sense that, you know, I'll take the opposite tack. What I don't need is 50 applications to do something. And, and, and every day I encounter something where I'm like, Jesus, I gotta log into this other thing over here just to pull that out so you can connect something to this.

And I think, you know, in the future, I think most people are just gonna wanna have an AI agent that pulls what they need, regardless of what app it's in. And all these apps just become headless services. You're never gonna have it one AI agent that does all you're look talking about a master agent.

I will, I will have a, a master butler that will talk to all these Other agents, the other agents don't. And that, and that, and that is something, Salesforce and ServiceNow and the Automation Anywhere or all that's the holy grail, right? This orchestrator.

But let me, let me take another tack on this. And again, I spoke about it on my shimmy says, we are definitely in some sort of m and a wonderland, right? We're seeing deals coming down the pike.

And every one of these deals have the word AI or agent ai, general ai, generative, A, they all have AI involved. And that's what's making it go. Now, is this a good thing or a bad thing?

Is it the sign of a healthy AI ecosystem in the natural order? Or is it the sign of a sick, there's some sickness here, there's some bubble or what have you, right? Some malaise.

And you like that word, right, Mike? Um, I do, I I'm reminded of Joe Kennedy getting out of the market. 'cause the shoe boy told him to buy stock.

Exactly. That's exactly it. That Wasn't, wasn't malaise.

What was a malaise? Jimmy Carter Carter though. Jimmy.

Jimmy. Yes. That was Jimmy Carter.

That was Jimmy Carter. That was Jimmy Carter word. It was God bless Jimmy Carter.

But here's the deal. I, The money being paid for these AI companies, whether they're really just an open source database company or something else, right? Is, is healthy, right?

You're paying a million dollars for a Postgres serverless. Postgres a billion dollars. Excuse me.

I don't know what the sale price was on, They didn't just, they didn't disclose it. Well, We'll find out. They're public.

Can I just mention, can I mention one thing? Sure. That it's kind of a pat on the back for the Futurum group.

Um, Salesforce was mentioning, again, the digital labor force in the digital, uh, labor market. And they cited the Futurum group, uh, report that came out that estimated it to be six. Wow.

$6 trillion by 2030 Seems pretty high. But, um, anyway, they're, they're using that as, as one of their, one of their, um, Points. Yeah, but I, I think we're seeing that pull back because I, I think what's happening there is people are realizing, and, and you know what, speaking of future, actually Daniel Newman wrote this on LinkedIn the other day too.

And I, I commented on it that at least between now and let's say 20 28, 20 29, what AI is gonna be much more of a co-pilot than a pilot. It's not gonna necessarily take away people's jobs. It, it's gonna augment people and make them more effective.

That's not to say that at some point in the future, it is gonna take people's jobs. It will, but not, not in the near term, right? Not in the near term.

It it's not, you know, for all those people running around saying it's gonna take everybody's jobs away, that's not happening right now. But again, is it a bit of a gold rush and people are trying to get in before we find out that it's the man behind the black curtain pulling the levers that is making all this Wizard of Oz stuff, right? And then, you know, if you're not one of the first companies that got in on this gold rush, you get some pretty meager rations, right?

You can start seeing fire sales and, and all of this stuff. We, we've seen this cycle before. It's boom bust.

And, um, right now we're certainly in the boom, the valuations are really high. If you've got a good, you don't even need a good business. But if you could slap an AI angle on it, one of these, So what motivates this Alan or Chairman?

What, what, what, what motivates this? Uh, I, I feel like this feels well, okay. Right?

But I, my sense, which is based on virtually nothing is Benioff is paling around with his, you know, VC, angel, whatever like that, the whole finance area. And he wants to be able to go and say, Hey, I bought this company. They're really cool.

They do all this stuff because it helps keep the money flowing. It's not marketing to, you know, the market customers, buyers, partners, it's marketing to the financial community will community. But I dunno, I dunno, you gotta look at this as a business guy.

Every company Benioff takes off. The, the board is one less company on the board, one less potential company that one of his competitors are gonna buy or that maybe will strike lightning in a bottle and become a, a real competitor to him. Right?

He's taking pieces off the board when you are the king or the queen on the chess board, right? And you have that advantage. You want, you don't mind trading Rooks and Knights and Bishops because you know, you've got the power, right?

As that board thins out, you, you have a strategic advantage. And, and he does. 'cause he was early in it, he's got lots of resources.

He's taking chips off the board. I'm gonna be deeply disappointed when that $6 trillion in, in GDP or whatever it is, only turns out to be a measly two or 3 million. I mean, come on.

Two or 3 million or a trillion. Another 1, 1, 1 thing about Mark Benioff, you know, God bless him. He's, I, I've known this guy forever.

If there's ever somebody who is going to hammer home something in terms of marketing, in terms of mind share, in terms of trying to oversell an idea, it's gonna be him. So that's why we're just gonna see an acceleration of a lot of AI related news. They also announced it on Thursday, last week, uh, Salesforce announced new pricing system around AI products.

So more to come. Yep. We, I, I, I think, you know, we're gonna see a deal a day here for a little bit, and then it'll be interesting to see, though, if the multiples stay high, right?

What, what the, what the, you know, sale prices are anyway, guy, I hear them calling your gate. Maybe we should just have a segment that's deal, deal of the day segment. I, we, I think we, we have the last three shows.

Yeah. Let's make a deal. I'll give you 50.

If you have a paper clip in your Tell, tell you what, next time I have the chance to, Alan, I'll, I'll, uh, I'll, I'll actually do a walking appearance and I'll get on the plane and sit down and my, That would be very cool. Yeah. Sort of.

Well, I gotta, Shane Moj Simpson run through the airport. Can I get an AI image of Monty Hall for this new segment? We could probably come up with that.

We come up With that. That's a good idea. And Jay, why do we have for, for our Guests, Right?

Um, I don't know how many people got that one. I know Mike and John did. Um, all right, let's call a, a wrap on this version of, uh, Textron Gang.

I, I'm, I'm, I'm afraid they revoked my, my, uh, my TSA pre or something. Um, we will be back tomorrow hopefully with even more great commentary on what's going on in our world. Most mostly about what's going on in tech in our world.

But until then, this is Alan Shimel for Text and Gang. Have a great day, everyone. Hey, everyone.

We're back here with what I think is gonna be our highlight of our coverage of the Imagine Conference here for Automation Anywhere in, uh, beautiful Conrad Resort in Orlando. I want to introduce you to Meher. Shukla.

Meher is the CEO co-founder of Automation Anywhere. He's doing this 20 years, right? For a lot of you working or watching this out there, you may not have even been working 20 years, let alone at one company, 20 years at my age.

I, I can relate, but Maha, welcome to Text Drug tv. Yes. Thanks for having Me.

It's my pleasure. So let's start right off with that. Yeah.

20 years ago. Yeah. Talk about the vision 20 years ago.

How's it, how's it changed? How's it stayed? The same, how do you, when you get outta bed every morning, what get what still gets you excited?

Yeah. That, that's a great place to start. Uh, from, from, from day one, today to today, the mission has been same.

What has changed is along the way, we had to invent few technologies. Some technologies got better in the industry overall, and as a result, we are able to achieve over mission better and better every day. And that mission was to reimagine how work happens.

Prior to Automation Anywhere, I had a chance to do four different multi-billion dollar journey. And by the time I had seen a huge part of the world, and I saw a part of the world where 70% of knowledge workers were sitting in a cubicles and doing work that you felt weren't the human jobs. And they could do so much better if you unleash human potential.

And I grew up in a small town in India, and from my own experience, I knew that talent is evenly distributed. Opportunities not. And the opportunity that was made available to me made many things possible for me.

So I didn't have to read a book to know that. Mm-hmm. It is my own life's experience, lived experience.

So the goal was to invent a set of technologies that take computers to the next level, and how automation and AI can do things. And in doing so, the biggest bigger vision was could we reallocate intellectual capacity of the planet to more worthy causes? What if we don't have to process invoices and claims and few other things?

Could we be doing something more exciting, more fulfilling with our lives? And, uh, so with that objective, we started, and 20 years, more than 20 years later, here we are. Uh, I, I get up every day.

And how often you get to real help, help relocate, intellectual proper Absolutely. Capacity of the planet. You know, I always say almost not multi-billion, unfortunately, but I've done four or five, uh, venture startups founded, co-founded.

And I, I firmly believe that every founder in their heart Yeah. Believes that in some way what they're doing is making the world better. The vision you've enunciated here.

This, this is in a small way, this is in a big way. Yeah. Right?

Free us up from doing these repetitive, I wanna call 'em low value. Yeah. Or they're not low value.

You know, we, we interviewed one of your customers from a light net, a light, I believe they're here. Yeah. These are people who are processing people's benefits.

Yeah. Medical benefits, health claims. Yeah.

Mission critical life and death. That's right. They're doing so much with automation and, and, and automation Anywhere to speed that up, to allow people to go do higher value things.

And these claims can just get processed automatically. We've all been there. You submit a doctor's bill and you gotta wait 30, 45, 60 days for it to go through.

Not with automation. That's right. Now, of course, this past year you've been in business as long as I have.

This is sort of a, a high watermark, a a threshold year where we're starting to see Yeah. Maybe people's visions for what AI can be Yes. Become real.

Yes. This whole new thing. Uh, a agentic process information, a PA Yes.

Talk about how that's fundamentally changed, R-P-A-B-P-A and all of that. Yeah. And how it, frankly, it's fundamentally changed your company.

Yeah. Um, so, so the in general, my view, uh, having done technologies for many years is that technologies take a huge step change. So if you look at last many years, what we have been able to do is use, uh, RPA robotic process automation, document automation and, uh, task mining and various other capabilities as part of the automation platform.

What all of these CAP capabilities were able to do combine is they were often able to automate 40% of the processes, but the other 60%, uh, were left, uh, uh, still manual. Suddenly, with the power of generative AI coming in, you were now able to take 40, what was 40%, sometimes all the way to a hundred percent or 80%. So just a one new ingredient, completely change the value proposition of what is now possible, not by itself, but in combination of everything else that existed before.

And when you add this ingredient, it changes the game. To me, it's almost like if you, for people who are into cars, you know, they sell that STP, I dunno if you're familiar, STP gas added of you add it to your gasoline and it boosts the octane. That's right.

That's right. That's a good one. Good.

That's, that to me is what we've run here. You already had gasoline. Yeah.

But we just put in some extra octane that takes this thing off. And, and it's an important, it's an important, it, it, it, the, we can't underemphasize how important that is. Right.

This is this game changing kind of stuff, but yet there's more Wait, there's more. Right. We, people are talking about artificial general intelligence.

Yes. And I'm not here saying it's gonna be next year or five years or whatever. I, I'm, I don't play that yet.

Okay. Yeah. But you don't have to get that big a GI, if you will.

Yeah, yeah. To have an a GI sort of, uh, influence in what you are doing. Yeah.

Talk about that. The, the, I am I'm with you there. I'm not a big fan of word a GI by itself, because it's an abstract concept with, you have no idea what what that even means.

But I think if you define a general intelligence in context of specific purpose, so for example, if you're, if you're, if you, if you're an intelligence for a self-driving car, the, the, it's, it's very, very clear that this car have enough intelligence to drive itself, yes or no. Right? So similarly, what we are focused on is that can we develop enough general intelligence to do, do vast amount of work as we call it, for a knowledge worker?

So can we give, uh, one of our, uh, one of our software, uh, uh, uh, mortgage applications to process and claims to process and supply chain and tax and audits and vendors, and a vast amount of these things that can, you just give it to the system and it has enough understanding of how to do this now in define that way it looks possible to achieve it. And we recently announced that we took a, for a step closer to it, towards it. And it is amazing to see how fast that is moving.

And, um, you, you, you're never sure, but it, that, that day looks closer now than it looked a few years ago. It's, it's not, I I say the same thing about quantum. When I first started talking about quantum computing, I said, I'm not gonna be alive.

Yeah. By the time, but all of a sudden, I, I would not be surprised if quantum's here in 2028 or 2029 even. Yeah.

I think it's the same thing. Same thing here. Yeah.

Same thing. It it all of a sudden that horizon's gotten a lot close. Closer.

That's correct. So Meher, you know, we're sitting here at this conference, beautiful conference center. You could feel an excitement when you go down by the, the, uh, stages and in the area.

Yeah. For you, what are the big stories for imagine this year? What are, what's the big message?

I think this year, as you said, was very pivotal for us, and we announced three huge, uh, uh, uh, announcement, uh, that moves the category significantly forward. The first one was an announcement to a, a significantly expanded, uh, agent tick process automation platform. It's one unique element among about a hundred other features.

That unique element is a process reasoning engine. Yes. It's a very powerful capability in it.

It, it, it works the way enterprise is needed to work, which is in context of my enterprise, in context of my past information, in context of regulatory environment, reason what I should do at this point in time. And that process reasoning engine is what every enterprise's needs. And it, it, it is the engine that will power all enterprise processes.

And we announced the capability of how far this reasoning engine has come and how transformative it would be for work. The second announcement we made was an availability of agent solutions. Uh, these solutions are not like your typical solutions and applications that you see that earlier.

We have seen for, uh, 20, 30 years. We frankly don't like some of them because you have to often do 20 clicks to get something done, and it feel like they're designed for an era 20 years ago. I think there is an opportunity here to reimagine how, how this work gets done.

It can, can you have a solution or application that is agent first case, autonomous first, and design with a very different mindset. So today we announced four different solutions and many more to follow, including many would be offered by our partners. So, um, we are looking forward to how that transforms the industry and accelerates the journey.

Um, the third announcement we made is about a very unique set of capabilities in most cutting edge of the products and our services and partner services to create a offering called Autonomous Enterprise. Now, this is, this is designed not, not necessarily for every customer, but we have customers who come to us and say, I want to get to this vision of how, how future, how, how, what future of the work is. And I wanna get to a point where 70% of all my current work is either fully autonomous or assisted.

Give me a choice to get there in the fastest way possible. I don't want to take the long route. I want to, I, I wanna take a flight and get there.

And so we, we, we, we offered with our partners a uh, combining many capabilities and, uh, we take this to those customers to take the extra, the journey even faster. So those are the three announcements. Excellent.

Two more areas I wanted to touch on with you. One is, as, as, as we enter this age of Agen ai Yeah. You know, we've seen announcements from Salesforce and, and ServiceNow and, and even some of the big hyperscalers Yeah.

Everyone recognizes that we're gonna be deploying multiple agents, maybe dozens if not more. Sure. This makes for chaos.

Yeah. Right? Yeah.

You need an orchestrator. You need, you need a, a commander, a master agent. Yeah.

Everyone has their own name for this. Yeah. Now I know, uh, automation Anywhere recognizes this.

You're also working with an orchestrator type of layer Yeah. Strategy. Yeah.

Would you expand on, without getting too far in the weeds, but would you expand on the need for the orchestrator? Yeah. And why you think your company is, is in the, a good place to be the provider?

Yeah. I think it is important. You, you mentioned that there will be many, many agents, but in our view, there are three types of agents, uh, that the first type is personal productivity agents.

And maybe many of the viewers understand that they're using one thing or the other. Today, the second types of agents are, we call captive agents. They're specific to certain application platforms like Salesforce and ServiceNow and others.

And there will be many others. But the third type of agents you call the super agent or the commander agent, that is the trick. That is the power.

That is where the power lies. And Automation Anywhere is going to own that piece, uh, to orchestrate the, all the, all different types of agents across multiple application and making sure work gets done end to end. Uh, it is important for our, for our vision to, to, to, to, to realize is that we have to break silos.

We have been working in the way since a post World War II era where everything is since siloed. And that is, that is how, that is why the work is as, uh, ineffective and boring as it has been. Um, there is an opportunity to change it with Automation Anywhere, orchestration, power, and some of our inherent capabilities.

And, uh, we, we showcase that imagine an ability to perform and orchestrate this work almost at a speed that you can give a command to and can work happen at that p that pace. And yes, it is possible for work to happen at that pace if you use automation. Excellent.

And, and that the announcements from Imagine this available now or coming soon, or, Uh, every single thing we showcased, we, we, with actual product, we showcased it. Uh, many, most of them are generally available. Uh, couple of, uh, capabilities will be available, generally available soon, but they were all available in beta and customers had a chance to play with it.

And, uh, that was very gratifying to see as well. Another area I wanted to touch on with, you know, I speak to a lot of CEOs, big companies, startup companies, everybody wants to be a platform. No one wants to be a product anymore.

Products aren't good enough. All of a sudden I gotta be a platform. Yeah.

But you need products that work on a platform. Yeah. Why is Automation Anywhere the platform?

You know, you hear talking that as many customers, they're are partners here and, and you know, people who are doing products that work in automation, the professional services that work with it. Talk to me about the vision for the platform. Yeah.

I think I, what wanted do is make sure I mentioned this earlier, but we, we, we are, we, we are the best in class platform company and always have been. But I mentioned earlier that we are, we have announced many solutions and many more to come. So on top of being the best in class platform company that it can use, we are now this new era of agent solutions that a business can use to, so make sure we are serving our customers, IT and business both through two different, uh, offerings.

Um, uh, if you, if you think about, uh, for, for a minute, think three or five years out, because sometimes you can debate, uh, you know, future in three months or six, but it is probably all of us could agree what it looks like. Multi R multi r The work is not going to happen the way it happens today. You are going to use, in my view, multiple LLM providers.

Yes. Because no one provider can provide you everything. It's perfect.

Right. You are going to orchestrate work across multiple applications. You are going to lead, go to the world that is as autonomous as possible and reduce reliance on manual processes.

That's the goal we can all agree on. Right? Agreed.

You could de depend how much, but as far as you can go, you are going to do that. Um, and, and, and, and, and you are going to need something that can execute mission critical processes. This is not about, uh, summarizing email.

That that's, that's not what everybody, you, you process mortgages and claims and supply chain and you assemble planes with hundred thousand parts and mm-hmm. These are important complex work and you, you are going to need some help in doing this in a complex world with, uh, if you just mention this, if you just think about this four or five things and need to orchestrate this work, you're gonna need something for it. And that's something is Automation Anywhere.

It's anywhere and it's available today. Perfect. There you go.

I love it. Last question. 'cause I know you have to leave.

I appreciate your time. Spend two days here talking Yeah. To customers, product, other partners, analysts, press, boil it down.

What are you hearing from them? What, what's your take back from these people that you're gonna turn around and say, okay, this, this is where we may wanna double down. This is where we may have to change something.

What, you know, based upon the feedback, what, what have, what have you learned? I, I think a couple of things. One is that they are very excited about the products that they're using today and as a CEO of a product company.

And that's the first thing I want to hear, that we are delighting our customers and meeting their needs. Uh, when they heard our new vision and where we are going, they're super excited. They can't wait to try it.

Uh, they, they can see the possibilities. And, you know, I've seen this every year where customers, uh, innovation capability is just unbelievable. They, they can, they can use this product in a way that you have never imagined.

And they make a difference in the world we live in, in a way that, that, that, that is so gratifying for us to see. So to, to, to, to interact and collaborate with this customers across 18 different industries, some of the largest companies, uh, on the planet. And we can't wait to see what they will, will make of it.

Uh, one of the key challenge is how to, uh, how, how, how to, how to help everybody understand the potential of it and how to, how to bring this to life in every department, in every function, in hundreds of different problems that the organization has to solve and how can AI transform it. And so our goal is to help together with us and our partners to help customers do that better every day. Fantastic.

Meher, thank you so much. We appreciate you taking the time to talk with our audience. Continued success with Automation Anywhere.

Can't wait for next year's imagine, but I have a feeling between now and then, this is gonna be a lot happening and going on, and it'll be interesting to look at this next year. Yeah. In light of what is to come.

Uh, thanks for taking the time to talk to us. Pleasure. Uh, I'll mention one thing, Alan.

The, in a technology space, uh, ever since I've started my career, the technology always takes longer than you expected and sometimes lot longer to achieve the vision that you started. That Takes longer than the hype. That's correct.

Right. The hype is always out in front of it. What is unusual about the, what we are up to right now with Agent Tick process automation is, it is take, it is happening faster than we have imagined.

I've never seen anything week by week, week by week, day by day. And this is unusual even for us, for all. Imagine the Customer.

Yeah. Oh, I, I I can't, can you imagine if we have a few more minutes? Yeah.

But it's one of the biggest things, and I discussed this with the, with the a few people here today. There's still the human element here, and the human element here is trust. Yes.

If the human can't trust what this agent ai or general AI or any of 'em we're gonna do, they're not gonna let it do. Yes. Right.

And we're at this kind of awkward stage where humans are still learning to trust Yeah. What the AI's doing. Yeah.

And so we're almost pulling the reins back saying it's going too fast. I don't trust it yet. Yes.

But once we trust it, yes. With the trust has to be earned. So one of the things we observed is we, we, we, we let our customers observe how it is doing it and put human in the loop where, where they, they have a chance to validate it.

Mm-hmm. Often what we observe is that after a person observes it for let's say three days, they get bold because this thing keeps doing the same thing again and again. And it's not as much fun.

And Sometimes boring's Good. Yeah. Yeah.

But so very soon. So if one person might, after watching three days say, you know what? I trust this thing, let's move on.

Somebody might say, I'll watch it for three months and then I'll trust everybody does it at their own pace. Right. It is important for technology to earn that trust.

Absolutely. That's a great way to end this. Thank you.

Thank you. Hey, I hope you've enjoyed our coverage for Imagine, uh, 2025 here in Orlando. Many thanks to Automation Anywhere for having us here.

We'll be continuing this story throughout the year. 'cause these are, this is, this is where it's at right now, right? This is where the rubber meets the road with ai, agentic, AI and everything else.

This is Alan Shimel. We're out. Thank you.

Hey everyone, it's Alan Shimel from Techstrong. We're here today at the beautiful Conrad Resort in Orlando, Florida for the Automation Anywhere Imagine Conference. It's been an amazing two days.

You know, there's so much going on in this area. You may or may not know Automation Anywhere. They've been a leader in, in, uh, B-P-A-R-P-A for years and years and years.

Um, but the world's changing and, and, and so is Automation Anywhere. And, uh, we're gonna talk about that. I'd like to introduce you to Audi Ganti, right.

Chief Product Officer, CPO at Automation Anywhere. Audi, thank you so much for joining us today, and it's great to have you here. Thanks for having me, Alan.

So I mentioned your chief product Officer mm-hmm. But I always like to give the audience a sense of your journey. Yes.

Right. How did you come to be the Chief Product Officer here? So, Alan, I joined Automation Anywhere about little under four years ago, and I joined as Chief Product Officer.

So as part of that product, technology community, um, those are my areas of responsibility beyond, uh, everything else that goes in a, you know, in a startup in a company our size that's growing so fast. Um, so I've been here about three and a half odd years before that, I was at Salesforce for 15 years. Um, joined back when it was one big cloud, one product line.

Uh, SFA, uh, I've seen it through this various growth, uh, transitions and spikes and, uh, kind of the major growth areas to join when it was, you know, sub 500 million left in 2021 when it was around 27, 20 8 billion. So saw that growth curve always in products. Um, was in different product lines, really at the core.

I love building, launching, growing product lines. Like that's my DNA. Um, and then was, uh, you know, Ryan ran 1,000,000,005 business, um, in, at the end of my tenure at Salesforce and decided I wanna be in the automation market.

That's why I joined the leader, which is where I Am machine anywhere. So Audi, you know, to me that's a really interesting, uh, history coming from Salesforce to Automation Anywhere. 'cause these are two companies that are kind of leading the, the path forward in terms of AG agentic.

Mm-hmm. Agentic ai and using agent Salesforce has of course, made a huge pivot That's right. To, to this agent based thing.

I, I have to ask you, four years ago, you're looking at coming to Automation Anywhere for automation, and of course automation is, you know, in the tech worlds always considered a good thing, right? Yeah. Um, did you have any inkling at that point that, uh, the agentic piece of this would be such a, I mean, such a, a, a watershed event?

No. To be, uh, very indeed. Right.

Uh, four years ago, no. Uh, but I always knew the automation market was right. Really ripe for disruption.

Mm-hmm. Um, yes, we are, you know, a market maker for RPA robotic Process automation, and there are other sub-markets out there, but there's no, like, the technology had evolved, but with AI and the kind of the broader, uh, market moves that's happening across various sections, I did feel that automation, that's one of the reasons I joined automation anyway, was a market was gonna explode. Yeah.

And this was something that was clear to me even in 2021 when I was, you know, looking at, I spoke to Hir and team about joining Automation Anywhere. Uh, because all of, you know, every business needs automation. Absolutely.

Um, as much as you might say that, you know, our operational efficiency is you, especially now actually Operation efficiency is even more important, but even more so growth. How do you drive growth by investing in the business? And that's where automation comes in.

Um, we are always been in ai, uh, frankly, even RPA and kind of BPA, there's always been a lot of ai, more ai ML um, but definitely this entire market of agents and Gene AI before that is completely disrupted. Absolutely. To me, I think of it as, you know, they were selling automobiles before Henry Ford mm-hmm.

But the whole concept of the assembly line and the little automation, and then so what, you know, it made a very different Yes. For a very different car market, automobile market. Now, we've talked around the edges about Automation Anywhere and, and you know, their history, they were a leader in RPA, they've been a leader in, they're still a leader and still in RPA.

Exactly. Yeah. Um, but when, when, as you, as Chief Product Officer, when did you, you know, go to your peers at the exec team and say, well, have you gone to your peers, I guess is a better question and say, Hey, we, we have got to go all in on egen, or maybe not go all in on egen.

Yeah. So, um, we've had, so there's a timeline to this. Mm-hmm.

We were the first into Gen ai. Uh, in fact, we started working with OpenAI and some other companies before chat, GPT, and it's primarily to drive more automation use cases around doc, specifically around document processing. Mm-hmm.

Um, and because we found a way where we could, for example, uh, process unstructured documents, so this is like back in 2022 before all the craze of chat GPT and, you know, the, the explosion. Uh, and for the next six to eight months, and this is post at GPD, we, investors started investing more and more in that area because we saw amazing results. Uh, and now our, one of our fastest growing products is document automation.

You know, three x growth, uh, 65% of our documents, our process using generative ai, you know, it's, wow. And we are talking about single digits last year and now 65%. So just crazy kind of growth there.

But really last 18 months. So we are first on Gen ai, then we started thinking about, okay, how do we tune this focus, this automation outcomes? Because most of the gen AI you see out there a general purpose, you know, summarize an email or, you know, do something else.

Write A blog article. Write a blog article. For us, our customers are asking, well, how does this apply to me in my automation, uh, use cases, whether it is accounts payable or what it might, whatever it might be.

So our focus is, okay, what are the investments we should be making to use Gen AI and then AI agents, but tune it and focus it on process automation? And so that was kind of something we went all in, I'd say about 18 months ago. Um, we as a product and technology team, we've been working really closely with not only the hyperscalers, the AWS and Microsoft gcps of the world, but also startups.

We've been coming up, up and coming, like AI native companies. So we've been kind of ahead of the market in that sense and getting access to early technology in before it's available in the market. Uh, so long story short, I would say is we, we kind of very much all in into what we call agent process automation, which is kind of our version of Agent a I, uh, and Mihir and the broad exec team.

Frankly, the company is all into, um, it wasn't overnight, but, but I think the last 18 months we've shown we are first to market with a PA first to market with agents. We launched it nearly a year ago before agents are cool now, it's obviously everybody's talking about it. Uh, and obviously we'll talk more about some of the new announcements, but I think the first more advantage has really helped us.

Absolutely. Well, yeah. You were starting from a position of strength.

Yes. Right. And I just wanna make clear, a PA is a gentech process automation, and that's, that's kind of the new term here that we're using for this.

I will tell you, I'm not as an expert on automation, obviously, as you are, but I've always looked, especially at RPA and said, it's great, but I always felt there was another shooter drop. You know what I mean? That something like an AI agent would really just ignite this.

Um, you know, I, I remember my first kind of exposure to RPA, I met someone in a startup group. I think I was a judge at a pitch event or somebody, and, and they had an RPA document mm-hmm. Basically system that was really cool.

But it, I just felt it needed the agent to really make it more independent. Yeah. More, uh, expensive.

Yeah. Mm-hmm. And, and now when you look at this, it, it, I mean, certainly it's, it, it's just as you said, when you go from single digits to 65% one year Yeah, that's Right.

With the top line growing at three x. So it's, everything's growing. It's not So three x and I mean, you don't see that, right?

This is like, when cloud came out, we didn't see this kind of growth. This is the kind of growth we saw when maybe the internet first wine commercial, right? Yeah.

We truly believe this, the phase phase we're in, and I think this is not only in automation, but broadly in the market. This entire agent tech, um, gen ai, uh, push is like cloud or even probably like the internet. I think It's bigger than cloud.

It's more like the Internet, bigger cloud. Yes. Uh, and so it's gonna make come complete not only disrupt markets, but you're gonna have the next Googles and Facebooks and absolutely AWS is the world from this coming out 10 years from now.

And so we wanna be very focused on this. I agree with, you know, I, I try to tell people that I've been in technology for 35 years, a long time, and I've seen a lot of stuff come and go and come again. Um, there are some things that as a technology person get me excited the cloud, right?

But explaining to my mother-in-law what the cloud is, all she knows is her pictures were stored up there. Right. Then there are some things that just change civilization, the internet, the cell phone, right.

These things. It's not just my geeky friends. Right.

It's, and, and that's to me what, what we're dealing with here with something like, you know, agenda a PA and, and agents. So, howdy. If you don't mind, I'd like to maybe pivot a little bit now and, and talk about the Imagine Conference here.

It's been two days chock full of product announcements, partner success stories, customers success stories, learning. Mm-hmm. You know, I think people are still, uh, wrapping their, he their hands, their arms around agent process automation.

What it, what it's more than just a name change. Yeah. Right?

What, what really does it entail? If you wouldn't mind, for all of our audience out here who weren't able to come Yeah. Who weren't here.

Tell us what we missed. Give them a recap, let make sure we, we, you know, let them know what, what's, what happened here. Yeah.

So it's been an amazing, what, two and a half days. We still got, uh, the rest of today. Um, we, I I was kind of big on three, three main things.

First, we talked about the amazing momentum we're seeing with Agent Tick process automation. This is something we launched last year as a category. It's really a new category, an expansive category.

Um, and we had our adon, uh, we had KPMG. Um, we had Washington Post some major customers talking about how they're using agent process automation and really driving incredible operational efficiency, but organizational change and business change within their, uh, you know, large enterprises. Uh, and across, you know, manufacturing and talking about, um, uh, uh, different industries essentially.

So that's one big thing. And, uh, across, uh, even the sessions, you, we had a lot of customers talking about, you know, the business benefits and the gains they're seeing with agent across automation. The second big announcement was a set of new innovations.

And that as a Chief Product Officer I'm really excited about. First was around the process reasoning engine. And essentially the way to think about the process reasoning Engine, it's kind of a secret sauce.

It's, um, our unique IP that helps us drive differentiated process automation outcomes in the world of agents, because you have general purpose agents. Our focus is on driving process automation outcomes. And the way we do that is, you know, we have our own unique models built our tuned on, you know, hundreds of millions of automation runs on our platform.

We have customer specific context, which name is personalization for specific customers, because, you know, customer manufacturing is gonna be very different customer financial services. It needs to be tuned towards them. Uh, and then these more, uh, goal oriented self-reflective agents, kind of cognitive AI agents that can, for example, um, pro, you know, look at unstructured content like a product catalog and make decisions on what's the right product, uh, replacement for a specific customer contract.

So there are different use cases, obviously. Uh, and so that entire process, reasoning engine ability to then, um, access in, in a very expansive enterprise tool set. So we're an automation company.

Customers use us to build RPA bots, APIs, process documents, build AI agents, bunch of different things. We also announced, so that pre was a big part of it. And, and underlying pre is also having an open ecosystem, open platforms.

We announced partnerships with the AWS, uh, they were on stage as was Google Cloud, uh, part where we're, um, you know, par partnering with Google Cloud, for example, on the A two A protocol. Uh, we've been partnering with philanthropic, um, model context protocol. So really having, kind of really focusing as an automation vendor on an, an open platform, an open ecosystem that's critical.

So that's all part of the process reasoning engine. We are the first and the only automation, uh, vendor to have that. And we believe that's critical for any company to drive differentiate process auto outcomes.

The second big announcement was general availability of our agent take orchestration engine. So we truly believe one of the core beliefs around a PA is it's a combination of deterministic and cog and cognitive. But if cognitive is driven by the process reasoning engine agent, take, our agent orchestration engine is a major, uh, investment in our deterministic processes.

The ability to basically run long running mission critical processes with the right governance and operational, uh, visibility. So that's gone ga um, with, with this summer. And the third big announcement we made is, again, in partnership with AWS, uh, around human agent collaboration.

So one of the things that in this world of agents, what's completely changing is how you and I engage with agents. It's not gonna be the same forms and traditional user interfaces. It's not gonna be those apps, you know, where, you know, like a, frankly a Salesforce or a ServiceNow, any of those.

It's gonna be these more collaboration experiences, conversational experiences, but where you get work done, it's not a chit-chatting experience. Right. Um, and so how do you translate user intent into process action, but again, personalized to that user, personalized to the employee, and personalized to the customer.

And that's what we announced with our partnership with Amazon Queue. Um, which is kind of under the umbrella of human. Yeah.

No, it's something we've covered a lot. That's right. You know, it's funny you mentioned Salesforce and ServiceNow.

One of the things we've seen is both of those companies, they don't want to just give you their agents. They want to be their agent orchestrator. Yeah.

You are the first, this is the first time I've heard someone refer to it as an agent orchestration system. But in my mind, that's always what was needed, sort of the Kubernetes is containers. You need an orchestrator for, for your agents.

And, but they may not call it that, but that's really where the race is, right? Everybody wants to, because we're gonna have a multitude of agents. We'll have an agent that interacts between you and I, between a process and I, between, you know, we're all gonna have a, a fleet of agents, whether they're ephemeral agents that are kind of single Yes.

And disappear. And I spin it on another one up next time, or persistent agent or however terminology you want to use, these agents are gonna need to be managed. Yep.

Do you see that as a key part of the Automation Anywhere value? Yes. Is managing this, this whole, this agent Army, if you will.

Yeah. The, the short answer is de a definite yes. Mm-hmm.

That's along with the process reasoning engine, which is kind of the secret sauce for all agent behavior. Mm-hmm. The orchestration of these agents is critical.

And we expect customers to build agents on our platform, on AWS, on GCP, on Microsoft Azure, on Salesforce, on ServiceNow, on open source platforms and anything, because frankly, you're gonna build agents for different purposes or different different platforms. And that's good. That's great.

Uh, but how do you orchestrate those agents and how do you also orchestrate both, uh, agents and deterministic workloads? Because when you extracting a document, you don't necessarily need an agent or you're calling SAP, you don't need an agent for that. You can just use an API to, it's Really, well, there's, there's no man's, that's a funny word to use for it.

No man's land, but there's this gray area, let's say better between what's an API call and what's the n agentic? And yous find a lot of, um, um, a little bit confusion in in the market. Yeah.

It is where everything is being called an agent. Uh, the way I see it is a cognitive task is an agent. And an agent can call a set of tools like an API, an RPA bot and other things to do.

Its work. But that's ultimate goal is you want to do something which is probabilistic. There are many things that you don't need a probabilistic, it's just, you know, you, you wanna call, uh, a Salesforce API an S-A-P-A-P-I or a legacy system, you need to, you know, uh, use RPA bot to access it.

You don't need an agent for that. So, but what we are focusing on, how do you orchestrate a process? And a process is gonna have deterministic steps, it's gonna have cognitive steps, and that's great.

And you're gonna have some of those cognitive steps like agents built on our platform, some built on other platforms, and we wanna own that orchestration layer. And like you said, the management layer. So how do you drive real type observability on how agents are operating?

Are they actually responding as you expect them to respond? 'cause again, they're probabilistic. So even 95% success rate may not be good enough.

It has to be 99 and more for certain mission critical processes. So having the orchestration layer and the observability layer is all part of our agentic orchestration system. I love it.

And, and I You're a hundred percent right. Make no mistake. The company that was orchestrating and managing these agents is in the catbird seat, right.

In the power seat to, as this whole thing starts to un unfold. We kind of stopped though, but let's go back to imagine. So that was two big Announcements.

So those are the two big announcements. And the third I would, uh, say is some major announcements with AWS. I already talked about some of the product announcements.

Yes. With Q with with q, uh, also with Bedrock and some of the other, uh, investments there, but also with Google Cloud are around, uh, partnerships on various products, but also on the agent to Asian protocol. Uh, I won't say March, but in Google's IO event, they're gonna announce some joint, uh, investments actually happening, uh, next week.

So, excellent. There'll be announcements there. So it's customer innovation and some key strategic partnerships.

I love it. And it's interesting, you, you've got Google, Microsoft, and AWS integrations here. Yes.

As one should if you wanna play in that cloud space. Right. Um, Haddi, let me ask you to look in your crystal ball now.

So this look, this has turned the, the RPA industry on its head. Mm-hmm. Two years from now.

Two years, 12 months to 24 months. How does this manifest itself? So one, the one caveat I would put in is, um, this entire agentic AI market is moving so fast.

I mean, two years ago, no, there was no consider agents. No. Now we only talk about agents.

Um, so I would put that caveat with this market every three to four months, it's just turning on its head. Right. Um, and I'm gonna put a crystal ball, like I gotta kind of guarantee that crystal ball.

But what I will say is I do expect, um, we've talked about some of the key, um, elements here, right? Agent orchestration agents being the probabilistic agents being um, um, you know, you can orchestrate but also control plethora of agents across different platforms. I do see there, there's gonna be a big focus on how do you stitch together these various agents for specific solutions?

How do you deliver value to the customer? Because right now it's still very much a technology talk. Mm-hmm.

Versus okay, what are the at scale deployments that can be out there? We already had some customers at the RDS and the of the world talk about how they're deploying it at scale. I think we need to go, today we have about 1500 live deployments.

I expect that to go into the many, many thousands, if not 10 thousands. And that's what I think the next two years. The the maturity curve Yeah.

Is gonna be important because ultimately it is. What we are in is how do we use agents for mission critical operations and for agents to be used in mission critical operations, it needs to mature, it Has to be mission Critical. Yes.

Yes. It has to, uh, it can, they can't be higher rates and there needs to be the right observability of the pieces. So our focus, and I think where the market is also gonna focus on a, across like our partners, the A W S's gcps of the world is how do we together partner to ensure our biggest cu all our customers can, um, roll out and kind of scale out these agent tech solutions and workflows at scale across all the business operations and see the impact on revenue, on operational efficiency, um, as well as on lower risk.

Because those are the three pieces, right? Revenue, cost risk on those three elements, we gotta see the impact of agents. And that's where I think the maturity curve is gonna happen for the next two, next two years.

But while we, while that happens, I think there's gonna be a lot of innovation going on in the market. We are just getting started on things like model context, protocol, agent to agent, uh, kind of, uh, protocols. Like how do different agents interact with each other?

'cause you, we all talked about how different vendors are gonna have different agents, so how do they talk to each other? That technology, we are just getting started. So I think there's gonna be a huge surge on innovation and commute, surge on innovation.

But I think what the market will demand is, well, how do I scale this algorithm my company and actually show business benefit? And that's gonna be our focus. Excellent.

Heidi, we're about outta time. I wish we could talk more, but we will because what, look, my feeling is we're at the beginning of the beginning of the Yes story, not even the end of the beginning. Mm-hmm.

So I look forward to continuing this conversation in the weeks and months ahead. But congratulations on a fantastic imagine event here. Thanks for coming on Techstrong and, uh, tech Strong TV and continued success.

A Thanks for having me. Appreciate it. My pleasure.

Alright. We will have more imagine coverage here from the Ooma Automation Anywhere conference in Orlando. Stay tuned.

Hello and welcome to the latest edition of the Techstrong AI video series. I'm your host, Mike Bazaar. Today we're talking with Sam Gong, who's vice president of Marketing for Work Span.

And we're talking about this whole issue that emerges. It's called the Partnering Tax. We all have experienced it.

We don't necessarily call it that, but once we get into it, you'll know exactly what we mean. Hey, Sam, welcome to the show. Thanks, Mike.

Happy to be here. So, partnering tax, we see this especially among big tech companies, but I think we see it everywhere. It's when a couple of companies get together two or more, and they kind of agree to work on something with a customer.

It involves them meeting each other. And all too often from the customer's perspective, it feels like all these folks are meeting it the first time when they come to their office. And it's not exactly a great outcome.

'cause nobody knows what the left hand or right hand or any other hand is doing. So Sam, how will AI help us get around this or avoid this situation? Because it does lead to a lot of lost productivity, for sure.

Yeah. Thanks Mike. And I think, uh, ai, AI is a part of the solution.

But if we think about how we got here for a second, this is a, this is a problem that I think has been coming for a long time. And if you go back all the way to the beginning, uh, we used to have verticalized businesses, right? You'd own the supply chain, you'd own the resources, you'd own manufacturing.

The goal for the business magnets of the last century was let's put it all in one stack and own all of it. And then we can control quality for the customer and we can control margins and prices. And I think, uh, the trend in our century has been for problems to get more complicated, for technology, to get more complicated.

And especially in the last 15 years. And now with ai, not only as part of how we work together, but as what we're offering to our, our customers, the pace of innovation has accelerated to the point where there is no one company in the world, not not Amazon or Google or Facebook or Meta or any of these companies that can, can own innovation at every layer of the stack. And, and so what that necessitates is, if I want to provide the best solution to my customer, I need partners to come in there with me.

I might need Nvidia for hardware and, and the the chip set that's gonna run the foundational l uh, LLMs. I might need Anthropic or Amazon Bedrock or Google Gemini, right? Like you've got piece together a solution that's best in breed for your customer, and that means that you're gonna work with partners.

And so now when you present that, that solution to customers, uh, the complexity tax comes in because your sales team, your marketing team, everybody that wants to position and describe that solution to customers, they're not only trying to keep up with what you're doing as a business and what you're building, they're trying to keep up with how is that current in the market? How does that leverage our partner's strengths? How have we put all these things together?

And, uh, it just gets stuck, uh, in a single person's brain. There's too much input to, to stay on top of all of it. So that's what we mean with the, with the complexity tax.

And, uh, of course this is, um, this is a really hard problem to solve. You're not gonna fix this with a SaaS application with if this, then, then that logic. Uh, you need cognitive tools, you need the, the benefits of, well, I can read a, a customer briefing from, from this partner and know how to digest that and boil that down so this salesperson can say the right thing at the right time on that deal.

So that's where we see the intersection of, of AI and the acceleration of innovation, creating this, this pressure to go to market with partners, whatever layer of the, the stack, whether you're a service provider or a software provider or an infrastructure provider. Uh, and then if you're an individual in that role showing up in that that room, like you said, uh, when partners come together, you don't want it to feel like this is happening for the first time. You want it to feel like there really is a joint value prop and a joint solution.

And we see AI having a big role to play in helping those partners collaborate as a, a unified front Your point, um, can I reduce the number of salespeople required for that meaning in the first place? 'cause I think that's also part of the problem. But if one salesperson might be able to, uh, accurately represent the others and then the others can go do something else more interesting.

Exactly. And I, and I don't think it's about reducing the number of people. I think it's about maximizing the leverage for people, right?

And so all the, the customers that work span talks to that are running sales orgs and partner orgs, uh, you want to provide the best experience you can to every buyer. You want to provide the most support you can to every partner. You never wanna say, oh, well that account's too little for us to, to take a look at.

Right? Or that partner's too little for us to take a look at. And so what AI helps with is you can take your people that have that subject matter expertise and they can divide and conquer, right?

We wanna put the individuals in the driver's seat and say, well, I can put 20% of my time into all these accounts if AI can help, help answer some of the basic questions and I can focus on these big deals or these big accounts. So it's, it's not about reduction as much as it is about leverage and being able to spread every person in their role to cover more deals and more accounts and more partners. And we'll also work.

Conversely, I assume that when there is an issue, it'll be easier to sort out, well, who might be the most likely one of this partnership who might be the issue or the root cause of set issue. And, uh, we can kinda cut down that whole customer service conversation. And, uh, exactly.

This isn't just a pre-sales problem. Pre-sales of course gets a lot of attention. Everyone wants to increase revenue and, and money coming in the door.

This is a full customer lifecycle problem, right? And so if you succeed in selling that joint solution, you have to have a plan to provide joint support for that solution. And that that doesn't just make the salesperson's job more complicated.

It makes the support team and the success team and the delivery team's job more complicated. So we, we say it's a complexity tax and we start with the initial sale, but it's the full customer lifecycle. Every single person that's involved in delivering that joint value to the customer needs this support in their role to be able to pull in those partner solutions and and deliver the joint success in today's market.

Will it become easier to onboard people onto these teams? 'cause I think one of the issues that a lot of organizations encounter is when they lose somebody from these teams, it's a, it's heart wrenching to them because it takes six months or more to replace them. Even if I, once I find the person, I, I think both things are true, right?

When you increase the leverage for each person in their role, that's a good thing. But it also means that potentially that person leaving means a lot of the knowledge walks out the door with them. And, and so what we're finding is, especially with partner managers, right?

You think about, I'm sitting at the intersection of all the complexity of my company and my company's products and structure and how we go to market, and then trying to unlock my partner's, uh, solution and, and the value they can bring, but understanding their complexity and product and go to market and all of that. You are, you are highly leveraged at that intersection. And it has not classically been a well documented process.

It's a lot of emails, it's a lot of, uh, PDFs and, and agreements that get shared back and forth in, in spreadsheets and PowerPoint. And uh, there, you, you can have a partner relationship management portal. You can have places where some of this stuff lives.

But that true understanding of how are we driving this partnership very much ends up in, in people's heads. And we're, we're, we're still early. We, I think what you're saying will come true, where as that person who sits at that intersection works more closely with ai, then that person will have flexibility.

Not if they want to change careers and go, go do something else or change companies, but also if they want to build another partnership and shift where they, they focus their time having AI act as that repository and their working buddy buddy with, uh, with an AI teammate that they're training that not only increases their leverage, but it increases their, their flexibility to spend their time on another partnership and give the business more ability to bring it in backfills or staff that roll with less of their time if they decide to do something different. Do you think as part of this, that the AI is essentially gonna function a little bit like the institutional memory? Can we all have people in the company who've been around for a long time and they serve that role, but um, you know, when they go, so goes the knowledge?

Yeah. I think, uh, institutional memory is a great way to put it, right? And, and if you look at any process today, I, I work in marketing and even in the, I haven't even been at work span two years, right?

But in my two years, I'm still uncovering institutional knowledge that wasn't part of my formal onboarding. It's not easily accessible to me. It's three clicks down in some folder hierarchy that I just haven't explored yet.

And we go to launch some program and I ask the team, have we done something like this before? And it takes a lot of digging to surface institutional knowledge. And, and so I think one of the things AI will do across functions, not just with partnering, is help people and businesses understand each other faster and, and shorten the onboarding and, uh, make it so you can provide value with more context sooner.

Because there's an AI that can answer your questions and do all that digging for you. How do I get started with this? I mean, do I just kinda like deploy some sort of AI agent and then it kinda monitors this until I get that level of institutional memory?

Or what is the length of time to get value in this whole approach? Uh, so with, with work span ai, we, we launched this last quarter and we've, we've been taking our first customers through onboarding. Now, uh, it really comes down to how accessible is your knowledge.

And if you have all of the questions and answers, and you've built really good field enablement programs, you can feed those field enablement, uh, uh, call recordings, training decks. The AI will pick those things up and then very quickly be able to disseminate that knowledge to the field. If your data's in good shape and you can feed it quickly, you get value very, very quickly.

It doesn't take a long time of passive listening to be able to provide value. And if you haven't had that, then you need to do some digging. You need to go back through your Slack threads and your emails, find the, the questions and answers so the AI can, can provide the answers that you would provide.

You need to do more cleanup if you haven't already organized that data. But then I think, um, and again, I'm speaking with my own experience using AI and marketing as well as our, our partners using AI in, uh, in their roles. You get into the habit very quickly where you say, oh, if I stay organized, I get this immediate benefit.

And I'm not an organized person by default. You know, I, I have a million sticky notes here on my desk. It takes me a lot of effort to stay organized.

And one thing that's really helped me is when I see that immediate return on here's what happens. If I can provide all these inputs to the AI, that then helps my team, the AI trains me to, to provide better inputs and stay organized because the benefits are so immediate. And so we've gotten great feedback from, uh, from our first partner managers that are training their AI and seeing that immediate, oh wow, I can, I can turn all of these questions that I used to have to go and answer on Slack over to my AI teammate now.

'cause I'm confident it, it'll say what I would've said in that situation and take those base level questions from my sales team and then I can go provide better support on the big swing deals that need to close this quarter. I think every salesperson probably has this experience where they have a call with a customer and then they hang up and about a few hours later they start kicking themselves and say, I should have said this, or I should have said that, or I shouldn't mentioned this. Well, the AI agent essentially remind them of things that they should say.

Is the conversations going well? Yeah, I think, um, across, across sales we're seeing this compression where as a seller, you wanna provide as much value in every touchpoint, right? You're an ambassador not just for your product, but for what it's like to work with your company.

And so tools like Zoom and Gong and all these call recorders, I think we're seeing what used to be a very long cycle of let's go back and let's do our QBR and look at our deals that we've won and lost and kind of do some, some introspection on why that happened. All of that is being surfaced into a much more real time immediate feedback. Let's dissect this call and have the AI look at your talk time.

io surfaces for every sales manager to coach their team on. Are you doing too much talking? Are you opening up the, the space for your prospects to tell you what, what they need?

And so, uh, with partnerships, same as with the rest of the sales cycle, that that kicking myself, uh, did I say the right thing? Uh, I think all of that coaching now with AI and workspaces, AI teammates is gonna be at the level where you're not just running better deal cycles, you're learning to run better deal cycles with a much faster feedback loop. So I think there's, there's two orders of acceleration here, right?

It's not just the availability of that information in the deal, it's also the availability of those best practices for sales managers to coach the team on, Hey, we could win more deals with our partners if we applied what worked over here on these other, other opportunities that we have in the pipeline. All right, folks, well, hey, sales has always been a game of confidence. And if you have an AI agent that's helping you figure out what your partners are doing and what your customer needs and what's going on in the industry, you're gonna be a lot more confident.

Sam, thanks for being on the show. Thank you so much, Mike. ai video series.

You can find this and others on our website, and we invite you to check them all out. Until then, we'll see you next time. Hey, everyone, we're back here at the Imagine Conference from Automation Anywhere, uh, in Orlando at the Conrad Resort.

It's, it's a beautiful resort and it's been a, a fantastic two days of learning here, I'd imagine. Our next guest is Micah Smith. Micah is VP of Developer Relations and Community.

Nailed it. That's it. You know, as I get older, it's harder for me to remember this.

That's, that's one question down, One question down. All right. They're not going to bing me on that one.

Anyway, Micah, welcome to Techstrong tv. It's a pleasure to have you on here. Um, we're gonna talk about a lot of things, but let's start off talking about you.

Yeah. So I'm best at that. Me too.

Um, imagine, so Micah, give people a sense, I gave them your title, but you know, titles are fungible today. That's right. Um, Give them a sense of who you are and what you do.

Yeah, so I lead the developer relations and community teams and Automation Anywhere. Mm-hmm. And for us, that means everything from our training, our Automation Anywhere University to our Pathfinder community, uh, that includes things like our Pathfinder framework, our AgTech Quest bot games, which are challenges that we create for developers.

Um, and basically a lot of community programs that we're running. So we just did a training camp for, uh, everyone to learn agentic process automation. We did 46,000 course completions in a total of wow weeks.

It was a huge accomplishment. Way more than we even expected, if I'm totally honest. Um, so I do that.

And our mission is really to empower automation programs to be successful, and that means empowering automation program leaders to know how to build and scale an automation program. And it also means training developers to learn how to develop, deliver, and deploy successful automations. Sure.

Let's unpack this a little bit. So you mentioned, I think it was called Automation Anywhere University. Yeah.

For those of the, our audience not familiar, give them a sense. What, what is the university about? Yeah.

So this is all of our efforts to train developers to learn how to build, deliver, and deploy, um, agentic process automation solutions. And so we've created a bunch of learning trails that are specific either to feature or specific to persona. So if you're a citizen developer and you have no background in development, you can take some of our courses and learn the basics, uh, the tenets of problem solving as we call them.

Mm-hmm. Uh, sequence selection and repetition that really forms the basis of every kind of problem solving. And those translate really well into developing and delivering automations.

We also have developer learning trails, so that if you are an experienced developer and you want to come to learn to build automations and processes and AI agents, we have training that you can take for that and, uh, learn to build solutions. One thing that I think sets our training apart from any other training that I've taken, maybe a little bias here, is we do this concept called automation skill stacking, where we'll teach distinct skills in several videos. So, hey, you're gonna learn about conditional statements.

Hey, you're gonna learn about logging. Hey, you're gonna learn about web scraping, right? We'll teach you those individual skills, and then we'll have a project at the end of that to pull all of those skills together in a build.

When you do that build, you're doing it in a timed and scored environment, which means we gamify it a little bit. Sure. You build an automation, you build it in this thing that we've called Agent Quest, where you're building it in that time and scored environment, and it gives you immediate feedback.

So you have the ability to go back and change it. Maybe you did it with web scraping before, now you want to do it with JavaScript. What does that mean for the accuracy?

What does that mean for the speed of processing? So we've kind of gamified the process of being able to learn and try different things in a safe environment where you're not gonna break anything. Like the worst thing that happens is it just doesn't work.

Start over. Okay. Yeah.

Yeah. I, I love it. This is a great thing.

So, you know, I, one of the companies I helped found over the years was DevOps Institute, where we were kind of the leader in DevOps training around the world of, I don't know, I forget how many tens or maybe a hundred thousand DevOps certifications. And what I came to realize is there are some people who learn for their own benefit, right? I wanna learn, so I know how to do this, it's gonna make me a better developer, it's gonna make me a better, a more valuable asset, more employable, what, what have you.

Mm-hmm. And then especially I see outside the us, um, there are people who are very tied into their certifications, right? I am a, they like to have all the initials after their name.

That's right. You know what I mean? Mm-hmm.

Uh, automation Anywhere University, can I get the initials or is it just for me to kind of better myself? Absolutely. We have an essential certification, which is kind of that base level.

We've got an advanced certification, and then we have an expert certification. And obviously those ramp up what's required of you. So you start with just being able to complete a, you know, multi-choice test.

And then it gets to the point where you actually have to build solutions and submit them, and we're scoring them to make sure that they're following best practices and stuff like that. Um, and that's something that's really important for us, uh, to make sure that not only are you going through these videos and learning, but you're actually able to apply it and you're being efficient, basically. Right.

I would assume continuing education's part of that charter or, Uh, yeah. Well, not formally. We, we definitely want to have continuing education for sure.

Uh, but it's not tied into universities as of right now. Got it. Excellent.

Um, you mentioned another term, Pathfinder. Yeah. Not everyone out here again is gonna know Pathfinder.

That's right. It Give us a clue what, what's that about? So Pathfinder is our umbrella for everything that involves our community, our learning, our training, and our materials and stuff like that.

We also have developed something called the Pathfinder Framework. And this Pathfinder framework is the nine dimensions that we think are extremely important for automation program leaders to consider when setting up and scaling an automation program. So you brought up a great point about training.

One of the things that an automation program leader needs to know is have something like a competency matrix developed, which says, I have the following people on my team, and they have the following skills. Does that help us to meet all of our needs that we have in our opportunity pipeline so that as we're churning through new automation and AI opportunities, we have the team and the skills that are able to deliver on that? And if we don't, then we need to have some training plans developed and stuff like that.

So those are the things that we're guiding automation program leaders on to say, Hey, do you have a strategy and vision? How are you thinking about governance? How are you thinking about your operating model?

Are you federated? Are you taking contributions from other people of your organization? Or is it just a centralized team?

How are you thinking about people and skills? Your opportunity and pipeline management, your development and deployment, best practices, change management, the way you report on metrics and the way you evangelize your program? All of those things are extremely important for an automation program leader to be successful.

It's one thing just to deliver code to production. If you're not able to tell people about the business value that is driving for your organization aligned to strategic priorities, you're missing the mark and you're selling yourself short. I agreed.

I I couldn't have said it better myself. Excellent. Um, a third thing you mentioned was around this whole agentic process automation.

Yeah. Which is, you know, in our last interview with Chief Product Officer Audi, it really is a pivot in the whole, you know, you look at the history of R-P-A-B-P-A, all, all of these things. Yeah.

Hindsight's always 2020, let me say. But in hindsight, you look at it and say, this was all great, but it was waiting for something like agen AI to come in and just ignite it, right? Yeah.

I think the most common misconception I hear about is people think that, oh, we did all this work in RPA, now we've gotta scrap all of that. No, we'll turn to apa. Right?

And I think the, the truth of it is, everything that you've been doing now becomes tools that are available to these agents that can be orchestrated as part of a complex multi-step process. And so it's really like, Hey, if you have a ton of those RPA automations that you've built, you have gold, right? Because you can use all of that with those agents, and you can enable them to do things like goal planning and execution, and they can determine in which order to execute those different automations.

I think of them as these are the programs that the agents are gonna run, unfortunately in, in the rest of the world, not in the RPA space, right? People make agents and then they've gotta figure out what are the programs? How do I, you know, what is this agent going to do?

And I gotta like kind of program. Yeah. Here we we're coming at is we already have all the steps.

Now I've got an agent that could go do all these steps for me. Right? That's right.

I don't have to worry about stitching together. That's right. The, the steps.

I think another important thing is you wanna have visibility in one platform where you've got all of these orchestrated processes with agents involved, right? And if you think about what you have with Salesforce and their agent force, and Workday has their own equivalent, those are great within those specific domains. But what if I had a place to pull them all together along with those RPA tools and all of that capability.

And so our platform enables you to work with third party agents as well as agents that are developed directly within our platform. What a great story. I want to talk a little bit about the Imagine Conference.

Yeah. It's been two days. There's a lot of customers here, a lot of partners, I would imagine.

A lot of Pathfinder members, a lot of that's right. University students. A lot of the people who you, they're your people.

Yeah. Yeah. Our community is out here strong.

How, how, what, what's been the big story of the conference for you? The energy around a PA and people starting to understand what's possible. And like we talked about, being able to use the stuff they've been building, but now do it in a different way, has been, um, definitely energizing and electrifying.

I think another thing that a lot of our customers are recognizing is that those opportunities that they had two or three years ago that they said, it doesn't meet our thresholds for ROI, or it's too complex for us to be able to do with this platform, they can now go look back at those opportunities and start to see like, Hey, we can actually do that now. And not only can we do it, we can do it really quickly and it's gonna be an awesome solution. So they're starting to recognize and realize that a lot of those things are great ways for them to find their first use cases.

I think another key thing is that people are getting hands on with building their own agents. We've got several sessions that we've been doing throughout the conference where people can get hands on building their own AI agents, like I talked about with that agent at Quest, where it's a time scored environment. We're teaching people how to do that.

The other thing that we're doing is teaching people about how to think about deterministic versus non-deterministic agentic workflows. So I've got a deterministic workflow, it's a bit more robotic, but it's still using a large language model to complete what it needs to do. I've also got a non-deterministic agent where I'm giving it that set of tools and letting it determine in which order it needs to execute based on the goal that I've given it.

I love it. Feedback from people. You know, there's been a lot of announcements.

Mm-hmm. And, and look at every conference like this, right? A lot of the announcements sometimes are forward looking, and then there's stuff, you know, rubber meets the road today.

What, what has been the, the community, you're, you're the voice of the community, right? What's been the community's kind of take on it? This will be a data-driven answer because we've done an exercise in our community lounge.

There's nothing less. Uh, we did an exercise in the community lounge where we've got this big board set up and we said, what's the feature that you're most excited about that we just announced? And what's the feature that you're most likely to implement first?

Okay. Right. Kind of two different things.

Automator ai, which is our ability to go from natural language to a built automation or process, is the number one thing that people are the most excited about. Right. Think of it as the, the vibe coding for automation.

It's a good way of looking at it. Automation co-pilot is the one that people think they will implement first, which is to say that's the one that's based on Amazon Queue for Business, where you're able to have natural language to interact with your automation repository with those assets, and be able to invoke automations, processes, API tasks, and, and let business users work with them. You know, that that's jives with, you know, text's part of Futuring group.

Lot of analysts working with that. And, and they recently did some, uh, my friend Mitchell actually did a whole report on agent AI for developers. Mm-hmm.

And I think that's the sentiment there. They're not looking for it to be the pilot. Right.

'cause a lot of people spend a lot of time saying, oh, it generates code. Is the code good? Is the code secure?

Is the code bad? They're not looking for it to be the pilot. They're looking for it to be the co-pilot.

And that, no pun with Microsofts or GitHub stuff or anything like that, but they want AI to be a copilot, not a pilot at this stage of the game. That may change. Yeah.

Right. As we become, as it gets better, as we become more confident and comfortable with it. But right now, I do think so.

It, it, you know, no brainer. That's the first thing they're gonna do, is take it as a copilot For sure. And that connects with all the training that we're doing.

We're suggesting that when you get that first pilot use case into production, your first use of an AI agent, you wanna have a human in the loop every single time. Absolutely. Because you wanna verify every single decision, every single extraction, every single classification that was done by those agents.

Yeah. You can start to peel that back later. But thankfully, our platform has that copilot interface, which enables you to have human in loop a hundred percent of the time.

And then you can determine programmatically if you wanna scale that back a little bit and not have every single one for review. And that way your users are focused on the exceptions or the outliers, rather than the normal use cases that come in every single day. Absolutely.

I wanna talk about another thing, if it's okay, Michael. So, you know, R-P-A-B-P-A-R-P-A more than BPA was kind of, uh, the engine driving the, uh, or, or one of the engines, the major engine driving the whole low code, no code mm-hmm. Citizen developer kind of movement, which is, look, it's been very successful before there was Gen ai and, you know, it burst on this data before this agenda.

Ai, of course, this changes the game, right? We're gonna go from, I don't know, 30, 40 million developers to maybe a half a billion developers mm-hmm. Because everyone could develop.

Yeah. What does, what does that mean for your chart? What does that mean?

Like all of a sudden, who is a developer to you? Well, I mean, I hadn't considered those numbers, but the weight of that all of a sudden feels, uh, overwhelming, Right? Yeah.

No, it's a 10 x. Uh, No. I, I, we're really excited about that because A, we've already got a great low code user interface that's web-based.

So anyone can just spin it up and use it. You don't have to worry about setting up your IDE or weird dependencies or stuff like that. It's easy to use.

It works in the browser. It works on Windows. It works on Mac os.

Wow. So we have support for that. I think the other thing is with Automator ai, like we just talked about, as long as someone understands their business problem, which is the specialty of business users, they'll be able to explain that and turn that into workable code and solutions that they're able to execute.

Citizen development has been kind of a contentious thing in the past, right? It has. Some people do it really well, some people haven't.

Personally, I've led a program where we were really successful with Citizen Development. Not every single person you train has gone on to build the most advanced automations in the world. But I think with Automator AI and where we're going with a lot of these new agent capabilities, it will enable more citizen developers to contribute in more meaningful ways than we've ever seen previously.

So, Micah, it has been a, a great conversation. You know, it's an exciting time to be having, especially the Automation Anywhere community You like, you did 46,000 people in four weeks, you know, on, just on Genix. So it's a great time to be involved in leaning and, and helping this community grow.

Keep it up, come back and keep us posted on this too. It's, it's a, an exciting thing. Um, that's gonna wrap up this one though, Micah.

Thank you very much. We are here at Automation Anywhere's Imagine Conference in Orlando at the Conrad. We're gonna be back with more state tuned.

Hey everyone, it's Alan Shimel. We're back here at the Conrad in Orlando. This is a beautiful, fairly, I think the hotel's less than two years old.

It's gorgeous. And it's the scene for the Imagine Con Conference from Automation Anywhere. And it's been a great conference, two days full of what we now call Agentic Process Automation.

It's really, to me, this is the coming out party where RPA has become a PA and we've been talking about it all day. We're gonna continue talking about it here with a, it's actually a customer of, of, uh, automation. Anyway, he's also been a key speaker at the event, but more than that, he's somewhat of an expert in this new field of ag agentic ai.

Let me introduce you. I hope I put, didn't put too much pressure on him. Let me introduce you to Rahul Patet.

That's right. That's me. Rahul is with a company called Alight.

Yep. We got that right. All right.

You're doing great. com. There you go.

All right. So Rahul, what do you do at Alight? So at Alight, um, so first, before I introduce myself, let me introduce Alight to you all.

Great. So, alight. Uh, at Alight, we help more than 35 million people to help access and manage their health, wealth, and wellbeing, uh, benefits.

Mm-hmm. And we do this across thousands of clients, right. Including 70% of the Fortune a hundred companies are our clients.

Wow. Okay. So think it's a huge scale, right.

And we have to do a lot Right. To help to ensure that these 35 million people get the right set of benefits for them. Okay.

I That's mission critical stuff right there. Yeah. And for a lot of these people, that's life and death kind of situations, you Know?

Yeah. And health and wealth are the two key ants. Right.

And then the wellbeing, all these three things. We take care of that. Me, uh, I joined Alight around three years back, and I lead the Automation Center of Excellence for that.

Right. They have an automation center. Yes.

They had it three years ago too. Yeah. Yeah.

Wow. So, so I, I, so when I joined Alight, I, at that time, there were three different platforms Alight had, right. Including Automation Anywhere.

And we, at that time, three years back, we were only focusing on automating back office task. Right. Which are all rule-based, determin mistake, those kinds of things.

And the other Well, that's what was possible then. Exactly. Right?

That's, that's what, and, and you're so, right. Right. That was the only thing which was possible.

There were no ai, AI and other things were just, just being started. Right. So that's what they were there, and we had 300 bots and more than a a hundred plus team, right.

Maintaining these bot. So I took over a large, complex and expensive portfolio. And, and that's where I decided to see, hey, can I, um, is there, is there a way I, I can optimize that?

And that's where I consolidated my entire automation platforms to a single one saying, and moved to Automation Anywhere. So you went from three, let's call them, uh, RR Rrp, three different RPA platforms to a single RRP to a single one. Or Automation.

Yeah. Automation T. Excellent.

Why did you, let me ask you, why, why'd you, why'd you settle on Automation Anywhere, if you don't mind me asking? Ah, Yeah, definitely. So they, we, we evaluated multiple things, right?

So we evaluate technology, we evaluated pricing, and we evaluated customer relationship, right? So what stood out, right? And actually all three stood out.

But the best thing that stood out was the customer relationship, right? They were, there was, and not that time, but today itself, right? To allow, right.

Even when we say, Hey, we need this, they are there right away, right? And then it, we are not a, a backlog for them, right? We are not any other company.

We are, they're, they're giving us the importance they need to from day, from day zero to till date, right? They're being giving us the importance. So it has been a great journey with them.

And that's the reason we chose it. The, the most critical thing was the customer relationship piece, right? Obviously, the cost also mattered, right?

And the technology mattered, but customer relationship was the great things. And I, I'm, I'm, I'm quite happy with it. It, it's so funny.

We're talking about things like robotic process automation and mag AI automation, but yet it's the human to human interaction. Human to human interac That matters. The trust, you know?

Yeah. The trust factor Lesson there, by the way. Yeah.

There's a lesson, right? No matter how fast and how good all the AI gets, we can't forget the humans. We will never be able to.

So somebody was asking me, and I, and I, I, in fact, I was talking to either IDC or Gartner, I forgot, right? One of them today morning. And they were saying, okay, we are, we are going edge intake, right?

What will happen to those people? Says it's not. That, right?

There is, there is a path. And I have, I've gone through this journey, right? Last year when I came over here, I was talking about a process which we have automated called Claims process automations, right?

That was a journey At that time. We had an, a huge number of people who were sitting in our back office operations, doing claims processing manually. We had to start that journey.

We started using ai, right? To either approve or deny a claim automatically. And today, I could safely say that, Hey, that bad job.

We, we process more than around 20 to 30,000, uh, claims every day manually. Right? Now, we are using AI to do that, and automation along with Automation Anywhere tools, right?

Uh, to automate that. And we are able to do, we, we are automating all of them, but our straight through processing rate that we are not touching and no human is touching, is around 30% of that. 9%, much better than what humans were doing earlier.

So it's, it's a, it's, it's a journey, right? That's what it is. And the new, the new use case, what we are doing right now, it's truly AgTech.

Right? That's, that's what I'll be talking about, right. In some more time.

Absolutely. Um, Raul, when did, in your mind, sort of the, the click happen, the eureka, that, hey, we're moving from RPA to Agen P, right? A PA, right?

Because clearly here it, imagine it's, it's a PA Yeah. It's a PA Last year it was RPA, and this year it is a PA. When did that click for you?

Uh, see, we are in this, right? We are in a era, right? Where things are changing so fast.

Yeah. Every 60, 90 days, you see some new LLM model coming up. I was talking to one other partner, and he's saying it's not about large language model, it's large action model now.

Right? So new words are being coined, right? We are talking about ai, right.

Somebody else, right? And I think me itself, he was talking about, uh, artificial general intelligence. A GI, yes.

A GI. So, so think about it, things that, right? We are right.

Last year, AI things are so drastically changed if we don't keep pace to that, right? And then it's not about that we'll be left behind, right? The customer end customer doesn't, doesn't matter to them.

Whether I use an ai, either I do it, uh, using a JI use, whether I'm using a RPA or an agent for them, the results are important. Till that time, the results are being shown correctly. Backend doesn't, it doesn't matter how, how we automate that, but it is critical to ensure that we give, see, for us, um, at Alight, our biggest challenge is, uh, or our biggest, our most critical period for Alight is, um, every year, uh, whenever the enrollments happens, right?

All of you, including you, you must be doing your Android enrollments every year. Sure. Every year.

But before, before that annual enrollment window opens for you, a, uh, we at highlight had to go and do lot of complex configurations Sure. For thousands of clients. And we have to do it in a very short span of time.

Right. And we need to ensure that whatever configurations we are doing, very specific to your own health, health and wealth, your, your organization's health and wealth beings, right? We have to do it correctly across thousands of them.

So we have an army of people who are, who's doing this testing is very critical. Right? And this testing automation, was there, RPO was there rule-based?

Was there it was not possible. We, we not, it's not that we did not tried automating this testing process, right. But it was never possible using the RPA, the kind of complexities that we have, the kind of variations we have across our customers, across the plants, right.

It's, it's humongous. Right? And it was, it was not easy to manage and automated testing script, right?

But because of this new agent, a KI, right? These new large models, which are coming up, right? So we are using a large language model for text, right?

To, to create test cases. And we are using another large language model, um, uh, and it's called the large action model that we call it, right? And we use the Automation Anywhere engine, right.

To create an agent A talks to an Agent B, and try and, and, and, and automating the thing. That's where it is. That's how it's not that I choose to use Agent A, right?

It, it naturally came just to answer that. Right. It naturally came that, Hey, No, that, that's, I, I, that's the promise of agent is it's not unnatural.

It's not, especially when we're talking about RPA. Yeah. It is the natural progression of, of taking, of Taking to the next level.

You Mentioned different LLMs. Yeah. Excuse me.

These are LLMs that you created yourselves. No, these are, we, these are all, uh, outta the box delivered by organizations. We are not.

So we are not leveraging any LLM that we are creating on our role. Right. We are.

And in all of These, but these are not the hyper, it's not OpenAI's, LLM or Anthropic or something like that. They are. Oh, they are.

Yeah. Okay. Yeah, they are.

Right. So we are using some of these organizations lm, right? Just from that.

So What do I think there's a word they're using for these now, not the farm or industrial. So there's a word for those commercial, these big Ls. That's what they are.

Man. I, today itself, I heard a new term, right. A GI and every time there Is a, well, a GI is the holy grail.

Yeah. Right, right. Yeah.

Yeah. When we get to a GI, it's like the singularity. Yeah.

You know, I, I, if I'm alive, we'll see what happens. But, um, but certainly when it comes to agentic and, and it comes to these, you know, commercial LLMs like that. But, you know, one of the nice things about agentic AI is you can rely on that, you know, large LLM large language module, but you could also really tailor more to your specific processes to your, to your world, if you will.

Yeah, It is. Right. And that's where the engine comes into place, right?

Yeah. That's where, see, the automation anywhere a PA engine comes into place. Yes.

Right? That's where the UI agent comes into place that I don't need to worry about what that is. 'cause the engine is working correctly.

Right. Then these workflow, that, so that's, that's how it's all integrated, right? It's all, it's all, we all have to work together towards that success.

Right? So there is some portion over here, some portion over here, right? There is a technology, there is a LLMs, right?

Where are AI and all those things are working together to make this agent AI process possible. Yep. Let me ask you a question, if you don't mind, Rahul.

So for most of us, right? We, we, you know, our, uh, benefit elections, you know, choice window opens usually towards the end of the year. Yeah.

Right? Sometime in November, Thanksgiving and closes by Christmas or, or three months making. Yeah.

As we sit here, it's may right of 2025, you already got through the 2024 rush. Mm-hmm. Right?

How do you think a PA is going to help you for the 20 25, 20 26 Exactly. The collection season. So, See, as I said, right?

And then it's a process. It's the business stakeholders, right? That have to, so these, my organization, business stakeholders, client people, they talk to, uh, our customers, right?

Those Fortune a hundred companies or those, all those, the thousand plus customers that we have, they have to go and say, okay, hey, customer A, what is the new, right? What is the new plan? Are you, are you planning to change some benefits that you are providing to your employees?

Right? We have to gather those requirements. The, the first cycle starts right after they have seen Right Once, once the enrollment in window ends, it takes couple of months, right?

For an organization, our customer, to figure out, okay, now this is the plan. I think these are the feedbacks they will get from their people. They'll choose, Hey, this is a newer plan.

They will choose. And that plan, whichever they're choosing, they will come and consult with our business teams, right? That says, okay, this is what we need to do.

So we get thousand, different thousand different customers, right? We get thousand different, more than thousand different requests beyond each customer has multiple requests coming in. And that we now start thinking of configuring into the system.

The agenda AI process will help us do the configurations, right? Um, and eventually we are not there yet, right? But yeah, we will automate the, read those requirements, configure it in our system, and then take out what is being shown in our, a aligned work life platform, right?

So to ensure that whatever the client has said is being displayed to them correctly. So that's, that's how the end-to-end process of an agent, a KI will work, and there will be multiples, uh, agents working over there. An agent, which might be configuring it, an agent might, might be executing it, and an agent, which might be validating it.

So, so it's an interesting journey over there. It's Going to be an, and this is a bring up another aspect of this ho agentic AI thing, and that is the idea. It's not one agent.

There's multiple agents. Some of them are kind of, you know, ephemeral single use. Some of them will be persistent.

Yep. One of the things I was, I was talking to the chief product officer at, uh, automation Anywhere you could watch the video. Audi.

Audi, yes. About the idea of orchestration Oh, yeah. Of agents.

Yeah. How mission critical is that to you, to using the multiple agents in your scenario? So in my testing process itself, right?

We have three different agents right now, and, and we are using an orchestrator over there, and that's what I was talking about, right? One agent right? Creates the reads, the requirements, which are, it's all in unstructured, right?

When the business people or the client team, teams of us talk to our customers, right? And take the requirements, just take it, right? And they document in a Word document, they put in an A table, it's all unstructured.

So they put the requirement in an unstructured format, right? The first agent could generates test cases and reads these unstructured data to say, Hey, this is what your test cases are, test scenarios for organization A will be, this will be how it'll be for B, and this is for C, right? That's how it works.

So that's one agent, right? The second agent actually then goes and execute these test cases in our a light work-life application, right? It does the execution autonomously, right?

We have this, and that's where this UI agent that we have, uh, the, uh, the feature which Automation Anywhere has given us, it's called the UI agent. That UI agent navigates through our enrollment flow for each of our clients and their complex scenarios, right? Autonomously, and then give them the reserve that's, that is, will help us in reducing our testing takes months, right?

If you ask how it'll help us, that testing, that months will be reduced to days because of stress agent take I and three different agents doing that. The third agent will actually validate it. But yeah, that's how it works.

And you also asked me that, hey, how your 2025 and 2026, I think even though I'll say it right, by, by by next quarter or two, we will automate 80% of the test state. 80%. 80%.

But it's not about technology. My agent a KI technology is great. It'll, it's not about us.

It's about the business, right. And our customers, whether the business will be able to, it's, it's about the mind shift change and a trust that should come in my business team's, business stakeholders to ensure a decision and an action which this agent is taking, is same as they were doing it manually. So I, even though I will complete 80%, but the adoption might be only 20% of I will be happy.

Like even if they adopt 20 to 30% of that, because it's not easy for them, right. And that, that trust factor comes forward. It's a journey.

I think the trust factor is the hardest part of this. Exactly. That trust factor will come slowly.

Yeah. And that should happen, right? It shouldn't, it's, the technology is there, but the trust will begin and it'll help, right?

Improvise what the agents are doing slowly. And I think that's, that's what, that's, that's the journey that will happen. Yeah.

That, that to me, that's not a, a question of the technology. That's a, just a question of human behavior. The human behavior, right?

How was today? How will you just, How fast do we want to trust this? Exactly.

Right. I, That's gonna be, and quite frankly, I think that's gonna be an interesting piece on the whole AI adoption curve. Yeah.

I, we were, we were talking to Micah Smith who runs the community program, developer community program here, and he was saying, you know, literally in one year, ai, you know, specific, specific uses of AI went from like single digits to 65% in one year. Yep. Because people started to trust it and they started using it.

It's the same thing here. I'll, I'll add one thing over there. Right?

Go ahead. Um, what I will say is, see, we are engineers, right? We are automation engineers, but the, our business stakeholders, they understand this process and complexities of our, uh, customers very well, right?

For, if we have thousand customers, we have thousand different business stakeholders for them and a team supporting that. We are a huge army of people supporting our customers. What we will do is we'll create this engine, this agent TKI platform of testing to them, and we'll give it to our customers, right?

So they can write their own testing prompts. It's an English thing tomorrow. Uh, I think down the line, these, the entire surface based automation that we are doing, it'll change, right?

It'll be a plain English that we will be doing and things will be happening. So that thing will go from us, from engineers to citizens back to our business to do those things. It, it'll be a great journey.

What we can think through that. Absolutely. Rahul, I want to thank you for coming on.

I know you were a little hesitant coming in here, but I, it sounds like you got comfortable. Okay. Yeah.

Thank you. It's a pleasure meeting you. Yeah, pleasure.

Okay. Thanks. Rahul.

Petit from, uh, alight here at Imagine Conference, uh, automation Anywhere in Orlando. We'll be back with more, As enterprises roll out production applications using AI model inferencing. They're finding that they are limited by the amount of memory that can be addressed by A GPU and a lot of the other architectural considerations.

This episode of utilizing tech features Steen Graham, founder of Metro ai, discussing Modern Rag and Agentic AI applications with Ace, Stryker and myself. Welcome to Utilizing Tech, the podcast about emerging technology from Tech Field Day, part of the Futurum Group. This season is presented by soy and focuses on AI at the edge and other advanced enterprise IT topics.

I'm your host, Steven Foskett, organizer of the Tech Field Day event series. And joining me today as my co-host from Soy is a striker, somebody you may recognize from our last season of utilizing Tech. Welcome to the show, ACE.

Thank you, Steven. I'm very excited to be back and, uh, we're here in beautiful Sunnyvale, California today. Thanks for having me.

Yeah. It's, uh, pretty cool that we were able to get together in person to record this episode. We're actually here for our AI infrastructure Field Day event.

And, uh, Heim is gonna be presenting, uh, this afternoon. And so we thought, thought that it would be fun to record an episode of utilizing Tech right here on the show. Uh, talk to us a little bit about, uh, what we're gonna be thinking about today.

Sure thing. Yeah. Well, the, the topic du jour is AI, as it has been for, uh, our last, uh, several conversations.

Uh, what we're getting into today is really around, uh, the inference side of the AI equation. So we've been spending a lot of, uh, calories there lately, uh, talking with partners, customers, understanding emerging use cases. Uh, as we've said, uh, there is no AI without data.

There is no data without infrastructure, right? And that's where soy comes in. Um, what we're learning is that the, the sheer magnitude of data, uh, on the inference side of things, uh, is just blowing up.

There's, um, uh, uh, that's not just our point of view. I mean, there's, there's analyst reports from McKinsey and Tech Insights and others that'll kind of reinforce that notion. But what we're seeing is people use these models so much and they, they, there's so much, uh, data involved in going in and outta these models during inference, uh, that it's really putting a strain on infrastructure and it's driving requirements higher and higher at a very fast rate.

And so I'm, I'm looking forward to getting into that topic a little bit today with our guest. Yeah. And, and, and importantly, uh, memory constraints pay a huge, or, uh, play a huge part.

So one of the things you're gonna be talking about today on both the podcast and as well as at the field day event, is how that can be, uh, reduced, uh, through some clever engineering. And speaking of clever engineering, that's why we've got Steen Graham here. Uh, Steen, welcome to the show.

It's nice to have you. Well, thanks for having me. And Steen Graham, CEO of Metro ai, and we do a lot of work around building AI agents and, and also model and hardware performance evaluation as well.

So talk to us a little bit about, uh, well, I guess let's, let's just go right into it. Um, many companies are trying to deploy AI-based applications. Many companies are trying to build applications that incorporate, uh, enterprise data.

Uh, retrieval, augmented generation or RAG has been a huge topic, but what we're finding is that that can place some pretty extreme, uh, stresses on infrastructure and require quite a lot of memory in order to implement in the real world, right? Absolutely. And I think, you know, that the modern kind of transition from your historical CPU first data center to A GPU First Data Center has caught a lot of enterprises in this chasm of them making that transition.

Meanwhile, the pace of change in AI is presenting all these opportunities for them to move to these modern software stacks, but their existing infrastructure doesn't quite work for it yet today. And, you know, I think what we've been working on with the team at soddy is how do we kind of look at those scenarios and pave a pathway instead of going, you know, full GPU centric data center with all your infrastructure, how do we kind of give them a pathway and an affordable TCO optimized pathway into deploying those, those latest AI agents and agent rag stack for their their use cases? Hey, Steve, can we, just to start, uh, no doubt some of our audience, uh, is familiar with retrieval, augmented generation, what that is, but can you give us a quick summary, you know, how is that different from just selecting a foundation model off the shelf and, and plugging it in and starting to feed inputs into it?

What, what does RAG buy you and why are folks so interested in it? Yeah, so I think the, the, a simplified view of things is, you know, several years ago with the advent of transformer base large language models, we, you know, you would just use a model, serve it in a chat bot, and then the model would hallucinate. And so obviously for, you know, enterprise quality needs, that doesn't work.

And so the, the next kind of iteration of, of innovation was how do we actually put your data close to that large language model? Um, and notably what, sometimes we use a, a vector database or, or a graph database in some scenarios as well. And graph graph could be super useful for, for different different use cases.

Um, but you just basically take all your existing data and usually it would be like, you know, for a particular use case and the, the associated tribal knowledge associated with how that use case is solved. And you're pairing that, that vector database with the language model. So now the language model before it generates an answer is querying, um, you know, high fidelity, accurate domain specific information.

And that's really kind of the simplified view, you know, of, of the rag, you know, environment today. Now, what you'll hear in the year 2025 is everybody will say, this is the year of AI agents. Um, and, you know, I have, I've definitely been hearing that.

Yes. Yeah. Yeah.

And I think AI agents are kind of, you know, an extension of that evolution where we're actually now allowing the AI to query like, and use tools from the company API calls, you know, to the company's internal CRM system or their HR system, or your supply chain system, or your Jira tickets. And so now we can, we can do things like take that domain specific information that RAG already has, add an AGENTIC framework on top of it, and then extensively do that where you can actually create a digital worker that can get the job done while a human's not in the loop. And that's where, you know, people are really trying to look for, where's my 10 x, uh, ROI with ai?

Turns out it doesn't happen when a person's in, in a chat bot, co chatty, and even a rag based environment, traditional rag environments, have had that kind of chat bot type interface with your own data. So you're chatting with your own company's data for higher fidelity outcomes, no hallucinations, but you're still not getting the scalability of a digital worker that the AI agents will provide. Yeah.

And, and that's, I mean, I guess if you wanna talk metaphorically, I mean, RAG is a great idea, but essentially, I mean, it's, it's a, metaphorically, it's a librarian with a really great card catalog who can look up things and make sure that things are contained within the data set and that they're the right things and so on. You can validate data. Uh, there's a lot of things to love about it, but the problem is it needs to have that really big card catalog or vector database, and it needs to have this huge data set, and that can take up a lot of space.

And that's been, uh, something that's, I think, held this back, even though it sounds great. How do you have that encompass, you know, your, your company's entire corpus of data? How do you Yeah, Well, I think, I mean, most companies, you know, sit on, you know, terabytes or, or petabytes of data.

So like step one is, is just basically organizing that data and the high fidelity data. And there's the, like the last 10 years we've went on a data journey. So for the companies that have transitioned to leadership data lakes, you know, they're in a good position to be able to, you know, start vectorizing that data.

And most of the database companies and data lake platforms are now offering the opportunity to kind of vectorize data as well. So there's a lot of opportunity and pre-work that's already been done to put the AI models in a position to succeed. But we're still kind of probably in the position where you do want to nail a particular use case.

So, you know, curating that data set for the pro the domain specific problem you're trying to solve, um, you know, who's the business unit owner that wants to solve that particular problem is still incredibly important. So you don't want to just dump everything, um, you know, into a vector DB and start querying right away. You're probably not gonna to get the highest fidelity results.

And a lot of corporate data is, is greatly outdated, um, as well. So there's, there's obviously some curation you wanna do to set yourself up for success, but there's a lot of pre-work already already being done today to, to make that possible. But that is the most important part of the journey, having your data organized and ready to go.

There's no question there's a lot of data involved in the, in the stuff we're talking about. Um, I'm curious, when you look at the actual, uh, architecture that, that these models are running on, um, and that this rag data is sitting on, can you give us a sense of, um, you know, is this stuff typically done, uh, in, in memory? Is there a lot of, uh, storage involvement in, in real time as, for example, an enterprise is running, uh, you know, an inter uh, an inference workload and consulting, some kind of rag connected external data source?

Mm-hmm. Yeah, I think, I mean, maybe like just looking at kind of the, the models themselves, like, and where, where people are battling right now with GPUs, because we're centering the data center and all our applications around GPUs. 'cause they're the bottleneck, you know, so any rational kind of throughput analysis always says the most expensive component is where you want to have the bottleneck.

And so there's a lot of pressure on the GPUs right now. The trade-offs that the, you know, the companies making GPUs have is it's really, really costly to put a bunch of memory in the GPUs simultaneously. The larger the model, roughly, the better the performance, the more state-of-the-art model, um, that occurs.

So there's this really big challenge, you know, and in the GPU memory around fitting big, big large models in the GPU memory. Um, and so that's kind of the number one bottleneck that, that we all, we all face, um, in the market today. Now, when you start pairing that with, you know, rag based architecture, which usually we're running the Vector DB on the CPU, and then, you know, we're, we're bringing in all that, that vectorized data, the memory hierarchy kind of levels out a little bit more, you know, more like a traditional, uh, memory hierarchy that you would an anticipate as well.

Um, but yeah, the GPU constraints are happening and then you're pushing the workload now into more of a full application systematic software workload that's, that's driving more of the traditional, you know, storage memory as well. When you shift to a rag based architecture. And AI agents is more, extens is similarly extensible to that where, you know, you're, you're running a lot of application logic for that domain specific use case.

API calls that are all happening in more like traditional compute infrastructure that doesn't need to happen on the G-P-U-G-P-U is just focusing on serving that model performantly. And when you say GPU, I'll just point out too that, uh, when it comes to edge especially, but even, um, increasingly in data center AI and cloud ai, uh, it's other types of accelerators too. I mean, you know, there are, there are definitely acceleration engines out there that, uh, in, in many cases can provide, uh, better service than just a standard GPU, but they have the same constraints that you're talking about.

In fact, in many cases, those accelerators have even greater memory constraints. Yeah, you, you absolutely nailed it. If you look at the companies, um, that are doing a lot of innovation in AI accelerators mm-hmm.

And I use JS as kind of to, to cover the AI accelerator world almost interchangeably. Um, but the, those companies in many cases made probably decisions to, to save on bomb costs Yeah. And not have memory, um, you know, in their systems that have, they've driven them to make decisions on how they serve models, and then they have to paralyze model to serving these large models mm-hmm.

As well. And some of them great wrote great systematic software to do just that. But this is like a big challenge in the world today, especially as you, you apply larger models, but also you apply a chain of thought reasoning.

We start to like massively increase the amount of inference calls we're doing, um, by giving the model the ability to kind of think through things more. Um, you know, now you're, now you're really driving a, a significant workload and ultimately a high memory footprint too. Not to mention people are announcing like nearly unlimited context windows, like 1 million token context windows at all.

And, and wanting to kind of like make sure we sustain that over time, which, you know, the context window versus rag scenario is, is an, is another, another trade off to ACEs because with these massive context windows, you're almost getting rag in that mm-hmm. MLM application as well. Um, but the kind of the fidelity of an enterprise application, I think still still likes the separation of a rag environment.

A lot of those context windows are mm-hmm. Teed up more for consumer based applications at this point in time. Right.

But definitely muddies the waters quite a bit. Yeah. One of the things we've been hearing about a little more often, and call it the last nine months, is, um, approaches for, for grappling with the increasing amount of data involved in inference, we've seen, uh, storage vendors come to market and talk about approaches, for example, for offloading some of your rag data and, and accessing that directly from storage.

Uh, NVIDIA is just a GTC talking about the key value cash and, and approaches for, um, you know, placing that in storage as opposed to in memory, especially as you, uh, involve more complex models or longer interactions between models and that just grows and grows and grows. Is that a, uh, is that a feasible approach? Is that, is that something that folks should be thinking about as a, as a realistic solution to the problem of memory constraints as more and more data gets pulled into the pipeline?

Yeah, absolutely. I think, uh, you know, kind of one of the, one of the gifts that we have that I think is under underutilized today in the AI world, probably 'cause we're all focused on like this GP memory constraint is, um, you know, disk NN And what that allows us to do with the, the disk NN base optimizations we're allowed to offload workloads onto solid state drives that traditionally would be run in memory. And, um, you know, while the indexing time it takes a bit longer to index it, the net results is your queries per second and performance improved dramatically.

So that's, I think, a little hidden, you know, hack that you can use to lower the, the memory footprint. Um, That's a little, I wanna, I wanna dig into that one. 'cause that seems a little counterintuitive, right?

When you tell someone they can read some data from, from storage as opposed to from memory and you're actually seeing higher queries per second when you take that approach. Is that right? Yeah.

And That's, yeah. That sounds wild. Yeah.

Yeah. Well, I mean that's, I mean the, I think the work that's been done, um, within the dis a n working group, and obviously we're spending a little time on the indexing side doing some pre-processing and some optimizations there. But once you've kind of made that, that trade off, when you index that kind of one time-ish trade off when you're indexing, then you know, you've got a little bit of the algorithm based optimization that will give you that, uh, queries per second performance.

And it's not like a two x type differentiator, but it's like same level, uh, you know, level performance, you know, plus or minus. Um, it can even in some data sets be dramatically more. But, um, and then you're looking at same, same level recall accuracy.

So just at a trade off of indexing time. Wow. And so it's indexing time, not even capacity.

'cause I was thinking that it would be a trade off with capacity as well. 'cause the nice thing about storage is that you can have a lot more capacity than you can have with memory. Yeah, that's, that's a fair point.

You're definitely using more capacity with that implementation as well. Uh, but you know, that's capacity that, you know, as long as you're using like the, the state of the art, you know, like PCI based drives, like that's capacity usually have, you know, in the system, uh, in initially as well that you're using for other applications. Yeah.

And that, and that's where I want to go to too. So, um, the side effect of making things, and again, even if it wasn't faster, even if it was just not slower, that's still groundbreaking. Yeah.

Right. And the side effect is that you have much more capacity, and so you can deploy applications with much, much more data to support them than you could in memory, even if it was not the same level of performance. Right.

I mean, even if memory blew it away, you'd still run outta memory pretty quickly. And I mean, so, you know, we've talked about soy, uh, you know, you guys have, you know, very big drives, you know? Mm-hmm.

I mean, I remember the announcement of the 60 terabyte drives and of 120 terabyte drives. Uh, I don't think we're talking about having 120 terabytes available to a rag application right now, but are we Well, I think it's definitely, you know, in the scope, it really, I think it really depends on how much high value data and enterprise has. And if they've got, you know, a hundred terabytes of high value data that's for a domain specific application that improves the quality of the output, you know, it's definitely, you know, in the scope of deployability today.

That's wild. Um, and of course it doesn't just have to be one drive. I'm a storage nerd.

I mean, absolutely. I mean, most, most storage systems use multiple drives, but just the fact that we have that kind of capacity that could be made available to these applications is really, really shattering because there's just no situation in which you could have that kind of ram at an affordable price point if you really wanted to deploy an enterprise application with many terabytes of, of data, you just couldn't affordably. Yeah.

Yeah. So speaking of affordability, one other cool thing we've been, um, having fun with recently is because that kind of core problem that we've, that we've always seen about GPU memory footprint, we thought it would be really interesting to see if we can offload the actual large language model onto the SSD. So this is actually very unique and, um, you know, what, what we've done, And I've heard about people investigating that, that's a really cool idea.

Yeah. And there's some, there's some tools and technologies. Um, in this case we're using a, a feature in deep speed, which, you know, in many cases we use for training applications, but deep, deep speed has some capabilities around model offloading.

And so, uh, what we've done recently is we've taken a 70 billion parameter model. Mm-hmm. Um, which doesn't fit in like a L 40 S based Nvidia, GPU.

Um, and we've actually offloaded the model into solid state drives. And while you don't get the same performance, you know, you couldn't deploy that model at all. You know, and so it gives you kind of model capability based on offloading.

So for people that haven't refreshed all their infrastructure mm-hmm. Or they're waiting to get the latest and greatest GPUs, you can actually use this technique to use a bigger model, um, on a lower cost GPU by SSD offloading. So that's kind of a, a cool innovation.

And I think just like disc a and n has evolved over time and performance has improved over time, I think we'll see a level of innovation and performance improvement and SSD offloading as well, that it's gonna be, it's gonna warrant paying attention to, especially as we're, we're increasing the number of, you know, chain of thought reasoning and applications and all these inference calls are exploding right now. So at some point you have to look at affordability. Um, and that's, that's a great way to hit a different, totally different level of entry point on pricing.

Hmm. The, um, the model off way offload thing is, is really compelling to me. It's a really interesting, uh, idea.

And I wonder, like, my, my, my gut sense is that that may be interesting to folks particularly, uh, who have interest or, or needs to deploy, uh, AI solutions at the edge, because in a lot of cases we have, uh, more severe power constraints, space constraints. You may not be able to put the latest and greatest GPUs in your Edge servers. Right.

Um, do you see that as a, as a potential play for this where, hey, you can now run a 70 billion parameter model on A GPU running at potentially much less power than the GPU would've otherwise needed, and now we can, now we can take that AI to new edge environments? Yeah, it, it definitely meets that, that criteria that you look, when you look at the edge, you think about, okay, we're, we're constrained, you know, from power footprint. Um, usually there's a big latency requirement at the edge, um, but the existing infrastructure at the edge that that's lit legacy, typically Edge has a little bit more legacy infrastructure, so it kind of checks all those boxes as far as trade-offs you'd wanna make at the edge.

Now, it might not be a 70 billion parameter. That might be a technique you use on a 7 billion parameter, or if you're deploying on some really legacy infrastructure at the edge, it might be a 700 million parameter model. So I think that it scales down to kind of the, the right footprint, uh, for the edge as well.

I wouldn't, um, ignore kind of the enterprise, uh, cloud applications here as well, because what's happening with, um, with the transition from chatbots to rag to AI agents over time is AI agents are running autonomous of human intervention. Um, now you can always, you know, human in the loop it, but what we want our AI agents to do is they want to, we want them to be our digital workers that are working for us while we're asleep, we're hanging out with their family and enjoying life. And then we wanna come back in the morning the next day and see the output, all the reports and documents that, you know, the AI agent conducted for us while we were enjoying some great sleep and some great family time.

And that can be done on a batch based processing node. So we don't need to like, you know, get the most high performance, uh, GPU in that scenario. Um, we can kind of use what MacGyver, whatever we have available today, and, and leverage that and then, and deploy it as well for batch based workload.

So I think AI agents offer us a great opportunity to do some trade-offs in, in latency. So like real time tokens per second, little less important for AI agents, um, depending on the particular workload. Yeah, we've been hearing that as well.

And, uh, with future, um, and some of the research that we're doing, in fact, we're starting to see people talk about using, um, CPUs for, especially for agent ai, for the same reason, because it's sort of an asynchronous workload. Um, also because there's a proliferation of CPU cores. The CPU cores have a lot of specialized functions.

In many cases, they're actually getting specialized AI instructions, and because they have greater addressable, um, memory in many cases than GPUs or accelerators do. So CPUs can look increasingly attractive for this. And especially, and, and, and with many of the things that you're talking about, I could see a lot of that going hand in hand with this CPU trend as well, uh, wanting to use more storage instead of memory to reduce the overall bill of materials to deploy some of these agentic applications.

Because essentially, um, you know, you kind of take this to its logical conclusion. We could see SY systems running, um, agentic applications on conventional servers with, you know, a reasonable amount of memory and a reasonable CPU and a reasonable amount of storage, thanks to the fact that we now have capability to use that. Is this a vision that you would share?

I think yeah, absolutely. I think there's a lot of opportunity to take that historical data center architecture and run AI agents on it, whether it's CPU, and we actually have a, a number of AI agents that run a hundred percent on CPU u, new GPU required. Now that being said, I think some of the, you know, older GPUs, fabulous performance still, you know, for those type of workloads as well.

So I wouldn't, I wouldn't start transitioning a hundred percent to CPU in all case scenarios. But for those batch based workloads where you're, where you're fine with a little bit more latency, um, it definitely works. So those existing data centers, I don't think need to be totally retrofitted today in all scenarios, you know, for a GPU centric architecture, um, we can make use of them for, for deploying AI and AI agents.

Uh, one, one more question from Mete. Um, I'm curious, since you're expert on agents here, and we haven't had one before on the podcast, I want to pick your brain on this. Um, let's say you've got, you've got a model, you've connected it to some rag data, right?

And, and what it can do without identifying is that the right term without giving it agency, is it can, it can give you insights and advice, right? And then, and then when you give it agency, you're now connecting it to other tools and systems and allowing it to, to take actions on your behalf. Does the, does the act of giving that model agency have significant, um, uh, repercussions in terms of the amount of data generated or used?

Like you, there's a lot of data clearly involved in training a model. Mm-hmm. There's a lot of data in RAG, potentially, and then whatever systems you're connecting the model to have their own dataset, which presumably existed before the connection.

But are, is there a, is there a big impact to incremental data simply by virtue of making a model agentic? Yeah. Is that something you guys have looked at?

Yeah, I mean, it absolutely, I mean, we see this in, you know, our AI agents that have, you know, chain of thought reasoning. Um, and the more autonomy you give these agents, I mean, the, the human is the bottleneck in the scenario. So if you, the, the better you design the workflow, the more API calls that that agent can do, the more tools it can do, the more, you know, autonomy you can give it and problems it can solve, it just massively explodes the level of data, uh, being created.

I don't wanna like, characterize that data as synthetic data, but I would say, you know, kind of non-human generated data footprint mm-hmm. You know, is, is massive. And I think, you know, obviously I think we're, we're kind of at, at the era where the non-human data generated footprint is, is gonna be, you know, much greater than the human generated data footprint, um, based on giving the AI autonomy as well.

And also kind of the advent of modern robotic tools are very data intensive. So I think, I think that that transition point's gonna be a very interesting transition point. It's great for the storage business, so ASU should be happy.

Um, and it's also valuable synthetic data, right? And, and this is kind of what we're struggling with with AI right now, is, uh, you know, there's, you know, the open kind of web data we've certainly ran out of, as far as, you know, training models. So then there's a path for synthetic data.

There's more human labeling, more reinforcement learning, and now we've got this whole category of, you know, AI agents doing chain of thought reasoning with their data footprint. And so, you know, all, all of that kind of non-human generated data is gonna be really important for the fidelity in the future of AI models. Because if that's high quality data, if, if we're generating high quality workflows, then that'll be super helpful.

Obviously, if the workflows don't work or the AI model's failing, then maybe that will, you know, devalue that mm-hmm. That type of data relative to human generated data at this time. Right.

So it's kind of a really exciting to watch that play out. Cool. Yeah.

Yeah. It really is pretty interesting. What's, what's happening here.

And I, and I'm actually excited because, you know, generally people, uh, there's sort of a, a thought that, uh, AI applications require absolutely cutting edge, high-end hardware that's really expensive, that consumes tons and tons of power that just, you know, basically there's a lot of negatives around ai and, and many of these negatives are true, but the industry is absolutely working to address those challenges and those criticisms. And in many cases, we're going to see applications being deployed on much more, um, restricted or, uh, modest hardware with, at, at a much lower price point. You know, and, and I think that all of this means that this, this technology can have a bigger impact than we might've assumed, simply because it doesn't necessarily require the, the, the biggest, baddest, hottest, uh, hardware to run on.

You can run it, um, more approachable on, on more modest hardware. So all of these things, I think, go in that direction, and I think that that's a, that's a positive for all of us. So that, thank you so much for this conversation.

I guess, um, what last thing would you wanna leave our audience with? What's your summary of, of this message? I think, uh, for me, I think there's, there's a lot of ways to deploy ai.

And I think the, the innovations we've seen in the last year are, you know, on affordable deployment of AI are probably 10 x, um, what we saw in the last 10 years. I mean, it's just an incredible pace of change on driving affordable models. And I think, um, what's most important is, you know, designing the right business workflow, you know, for these autonomous workers or AI agents.

And then, you know, figuring out the deployment methodology. There's so much innovation happening on, on affordable off the shelf hardware, or even affordably deploying a state-of-the-art hardware, um, that I wouldn't let that get in the way of, you know, your company's innovation. Excellent.

Well, thank you so much for joining us. It's been great having you. Um, before we go, where can people continue the conversation with you?

ai. Excellent. And, uh, ACE, it's been nice seeing you again.

As I said, you were one of the co-hosts last season. Um, so check out utilizing Tech season seven. Where else can people catch up with you?

Where have you presented recently? Or, or where are you gonna be? Uh, boy, oh boy.

There's, there's, uh, a lot going on at Solid I these days. Certainly. com/ai, where, uh, we hope to be featuring some of Metro AI's excellent work in the near future.

Um, we'll be at, at conferences all summer long, so, so keep an eye out at, at all the big ones. Uh, for now my head is spinning, uh, with all the implications of what's seen is talking about. And so I need to go have a, have a lay down and kind of chew on some of this stuff.

But really, really appreciate you being here, Steve. Uh, I've learned a lot and, and thank you. Yeah.

And, uh, me as well. And, uh, I will point out that, uh, by the time you watch this episode, the solid IME and, uh, Metro AI presentation will be published on YouTube. Just go to YouTube and, and search for solid IM and Tech Field Day.

And you'll find that, uh, that was part of our AI Infrastructure Field Day event, which happened in April. Uh, we're also gonna be doing an AI Field Day event, a Cloud Field Day event, and we just announced another AI infrastructure event later in the year as well. So check out the Tech Field Day website for more information about that.

So thank you very much for listening to this episode of Utilizing Tech. Uh, you can find this podcast in your favorite podcast application, just search for utilizing tech, or you can find us on YouTube. If you enjoyed the discussion, please do give us a rating or review or a comment we'd love to hear from you.

Uh, this podcast was brought to you by Soy, as well as Tech Field Day, which is part of the Futurum Group. com. And we are present on the socials.

You'll find us on x Twitter, blue sky, and Mastodon. Uh, just search for utilizing tech. Thanks for listening, and we will see you next week.

Hey, everyone, it's Tim Cook in Hot Water with El Duchey. You're watching Textron Gang. Hey, everyone, happy Monday.

I hope you had a great weekend. It was a great weekend. I was up in Boston, actually.

My son, my oldest son graduated law school, and it was quite a, uh, well, as a parent, it was a proud moment for me. So all good there. Um, but we're back here on the gang, and we've got some great gang members.

Let me get right to them. First of all, joining us from some airport somewhere in the world. He is a future analyst, visible impact principal, analyst, Textron loyal Techron gang member, moving up the ranks here, the one and only Guy Currier.

Hey, guy. How are you? Uh, good to be here.

Yep, I'm good. I'm in the Orlando airport, the wonderful Orlando airport on the way back. 'cause I spent the weekend, uh, with my uncle who lives in here.

And, and, and also, I do wanna say my daughter is starting college, uh, next year, but, uh, her plans to go to Ross School, so congratulations to You. Thank you. You know, I mean, he did the hard work.

I just read the checks, but, um, nevertheless, very proud. It's a proud moment. All right, moving over from guy, one of our newest members of the gang, but she's been on three or four already.

She's a regular, excuse me, cybersecurity expert extraordinaire, Teri Robinson. Hey, Terry. Good to see you.

It's so good to see you. And I wanna extend my congratulations, uh, too for, uh, son's graduation. Uh, my daughter graduates, uh, today, Monday afternoon, uh, from law school as well.

So, um, Really good for you. Next Stop the bar, I guess, right? Yeah.

Yeah. I think well in bar. So he's taking the mass bar up in Boston, and it's the end of July.

Yeah, they're all around The same. That's when she's taking the New York bar. So, Yeah.

Very cool. Very cool indeed. All right, moving over to Silicon Valley where he looks, he looks downright Rosie in the cheeks today.

Even he, I don't know, maybe, I don't know if it's makeup or someone's been pinching them, but he's our Silicon Valley editor, Jon Swartz. Hey, John. How are you, man?

I'm good. I'm back at the friendly confines of Belmont. Congratulations to all our college graduates and people going into college.

Um, uh, it's great to be here. Absolutely. It's great to have you on.

And then finally, he might be graduating from going to minor league games to the big league. Well, Mike, you, you had the subway series this weekend, but, um, we're here not to talk baseball, but Juan Soto got booed really bad in it, and I gotta tell you the truth, I felt good. Um, our Chief Content Officer, Mike Vizard.

All right. We Got, we got two outta college, one in and one to go. So I'll be working for a while.

Yeah, That's what that means. Um, anyway, let, let's move into it. I, I kicked it off.

It seems our friends at Apple have, uh, have crossed the MAGA line. You know, um, they, well, they're not gonna, you know, they're not gonna build more iPhones in China, but they wanna build them in India instead. And, and, you know, uh, while he was in the UAE giving them 500,000 GPUs from Nvidia to build the biggest ai AI center there, he, he took the time out to chastise Apple and Tim Cook about building iPhones in India.

Mike, what do you think? Well, it seems like, um, to your point, we talked about the whole, uh, meddling in the GPU business last week, and now we're extending that out to smartphones. And he apparently told, uh, the CEO of Apple that, um, manufacturing in India was undermining the trade position for the United States government.

So all these things now are somehow or other bargaining chips. But guy, this is unprecedented in my mind, but what's your take? Well, I think what's unprecedented is the degree to which just lip flapping is being taken seriously because of the position of the person flapping lips.

I mean, I, I I, you know, last week, um, the Trump was out in, uh, in, in the Middle East being, you've heard the phrase that being treated like royalty, but in this case, take it literally, literally being treated as if he were a prince, um, or a king or what have you. Um, and, uh, I, I completely believe that he had the, the conversation with Tim Koch. I do not necessarily believe liter, I don't definitely don't believe literally what he said about that conversation.

Um, uh, other than the detail of saying, Hey, you know, you, uh, you have to manufacture everything here. Um, I think he probably did say that. Um, and, uh, apple, I mean, pitted the poor tech billionaire, CEO, right?

Um, who has founded, uh, in Tim Cook's case, a lot of that billionaire ship and the, the deserve success of Apple on manufacturing and assembly, uh, uh, iPhones, but also a lot of other devices in China, and is trying to find some way to find stability in supply chain, um, in the midst of complete instability in trade policy. And to actually, tariffs is the usual word, but all kinds of uncertainty over how much, uh, costs of manufacturing, assembly overseas are gonna be, right? So they move to a different, like, what would anybody do in that position, billionaire or not, they're gonna move to another source.

And they've increased the investment in India from, I mean, John, you wrote this story. I think they're moving towards something like 25% of iPhones being manufactured in India. Yeah, 2015, yeah.

25%. Yeah. Is that 15% a few years ago?

Um, years ago. But let's call this for what it is, which is, uh, another thing for, uh, a politician to say and take credit for that doesn't necessarily have any actual foundation. Be it our president happens to do that about 99% of the time, instead of 50% of the time, or 20% of the time, depend on how civically you feel.

I would not take this to the bank in any way, except for one thing. We still don't really know. We don't know what Tim Cook said.

We don't know what Apple's gonna do. All we have are the words of somebody who says all kinds of things all the time. And, and I, John, I really appreciated the way your story was liberal with the quotes of our president.

Um, instead of trying to take them at face value, just present them to everybody. Yeah, no, it's like a mob boss, you know, trying to strong arm somebody and shake 'em down. I mean, this is, this is to think of, I think of the absurdity, and I know Alan's gonna weigh in on this.

We've got maybe arguably the greatest CEO in terms of operations and running a company who's trying to do the best thing for his shareholders and, and his employee and customers. I'm Employee, right? And customers too.

He's trying to diversify the, and customers, especially customers. He's trying to diversify the product line so he doesn't jack up the price of the product. Uh, he's keeping his options open.

And in fact, from what I understand, within Apple, they have, um, plans in place to either move harder to India or have to, or fall back into China. What, what it must must be so galling to cook is this idea that Trump is gonna probably change his mind yet again. I mean, he dropped the tariffs against China.

He doesn't change his mind. John, John, he doesn't change his mind. These people take a stick and hit him on the head and say, stupid back down.

He pulls up like a cheap suit. I'm you on the, but here's my, here's my Report. It doesn't change his mind.

But for, for for Apple, the fear, the fear is they have this exemption in place, right? He's gonna lift that probably, I mean, he's gonna change his mind. Yet again, it's a guy who, who, who doesn't understand basic math trying, telling somebody who's a genius at operations.

I'm just gonna play games with you, despite your commitment to spend hundreds of billions of dollars on the US over the, over four years in manufacturing, what have you. It, I mean, the whole thing is so galling and it must be chilling to any other tech company or, or any company that has to deal with this, this bs. And it's Going, I think, a shakedown, I think your essential point is it's a shakedown.

And the, and, and the, the, the important thing is that the shakedowns never stop. They never stop. So if you give in once, Here's the important, it doesn't mean they'll stop the, it's not just a shakedown.

It's not just, it's a shakedown look, should Tim turn around and buy him a nice plane, and then everything will be okay? Right? Is that what it takes?

Because, and you know what, for with all due respect, Camilla Harris brought this up in the debate with him, the one debate, 'cause he was afraid to do more 'cause she ran around him. That foreign leaders know his game, they know flatter him or flat out bribe him, and he lays down like a puppy dog to scratch my belly. Right?

That's what you're dealing with here. But let me, let me sit down, let me unpack this for you. Besides buying him a plane or something that'll mollify this guy and let you do anything you want.

Let, let's look at what it is. This, this. SOB is in the Middle East cavorting with people who have financially supported terrorists, who have blown up stuff and killed people in the us, let alone in Israel and everywhere else, right?

It's a known fact that the Qatari support Hamas. It's a known fact where Bin Laden's financial support came from, right? But also, okay.

And he's going to use Apple as a whipping boy, a fine corporation that's already pledged $500 billion. He's going to use them as a whipping boy while he allows UAE to, to, as I mentioned, get 500,000 Nvidia GPUs to build an AI campus. They're perhaps the biggest in the world, while him and his sons are making $400 billion deals to buy golf, to build golf courses and build buildings over there, right?

Didn't think it was important enough to go up to Turkey and deal with anything around this Ukraine, Russian war. 'cause that's really just lip service until Vlad pulls his string and tells him, get over here. Right?

This whole thing is ridiculous. But let me give you another thing. Apple's duty is not to the United States of America, though they're a US company or a US based company.

They have customers and interest around the world. Tim Cook has one boss, and it ain't Donald Trump. It's the board of directors of Apple and their shareholders.

And he has a fiduciary duty to do what's best for Apple and their shareholders, not even their customers, their shareholders. And that may or may not be in sync with what, what's best for us policy. But this is what happens when you have global corporations.

There's a, and it's high time. We remember corporations have a duty to their shareholders. If Tim Cook starts doing what's right for the US at the, at the cost of his shareholders, he's going to get fired and sued at and right for, so imagine, Imagine if tr if, imagine if Trump had his way.

So say Apple and, and in, in, in just in she panic was forced into doing this. It's, it's, it's absurd. It would take years, first of all, to ramp up production.

It would cost 30 realistic For an, but it's like, but here's the visual. You got this 78-year-old spelt 280 pound person with Jensen Wong, Elon Musk, and all these other syco fonts glued onto his hip or his backside or whatever, and he's telling him, yeah, give him 500,000 GPUs. Give them some star link.

I want my name on that building. Give me this plane. Dude, did you ever watch the Star Trek episode where they land on the planet, that somehow there was a book like Al Capone's Chicago was left there, and 500 years later, the whole civilization is built around the gangs of, of Chicago.

This is what we live in. This is what we live in. It's, it's, it's ludicrous.

It's psychotic. Stop, stop the nonsense. Hey, I have An idea.

Maybe, maybe Apple. Maybe Apple, maybe Apple should build a special Trump iPhone like they did for you too. Maybe they, they, they appeal to his ego and narcissism and they build a, a trump A key on this.

No. I'll give you a better example. Do you remember Godfather part two, Michael Corleone and Hyman Roth go to visit the, the president of Cuba, uh, Duarte, right?

I think, no, that, that was a different Banana Republic, whoever was the president before Fidel took Batista and Batista, they're sitting around these table with all these captains of industry, much like TR is, and he passes around a solid gold telephone as a gift to Batista from it, from ITT. And everyone look around and everyone looks at the phone and they pass it around. And it's very, it's the, this is the same.

It, it's like out of a frigging movie, dude. Come on. This is, we're living, we're living in a, in a nightmare.

We live in a nightmare of what the US should and could be. That's the bottom line. You know?

You know what was really also telling Alan is when you mentioned the sink offense, we had the, the president of Alphabet. I was, I-B-M-C-E-O, Sam Altman was there, Qualcomm's, CEOI mean, it was, You don't think it's the same thing when Putin comes somewhere and he brings gas prom and that prom and the other prom, and they all go to a prom. It's the same, it's the same thing.

It's the same thing. I, you know, I'm, I'm not, I'm not, I'm not really sure though. You're characterizing the nightmare correcting.

I think there's a lot nightmarish about this. But in it, in the context of this show, this podcast, the Nightmare is that there are businesses and individuals around the world who rely on Apple and the iPhone in particular to conduct business or to live their lives. And they have no idea.

It's not that they think the price is gonna go up or down or whatever. They, they, that whole platform for them has become less reliable in terms of cost use, what have you. And that may be one piece of it.

When you think about the fact that these kind of, this flapping of the lips can happen anywhere with any company or country randomly, and continue to create that uncertainty. We don't even have to talk about business uncertainty. We can just talk about the uncertainty of individuals around the world being able to do the things that they wanna do.

That I think with technology in the, in the context of this show, so the rest of the nightmare, you're talking about having a kleptocratic or godfather as a president, or whatever it is. I agree with you, that's a nightmare. That's a political nightmare.

It's a global economic and global political nightmare. But I, I always try to bring it back to the fact that this is text drop. And I would like people to understand like, what do I do now?

Great. The flat slips, what do I do? And I think it's just recognize that the problem you're facing, you, well, I buy, you know what to begin with.

Recognize what the problem is within this context of technology planning is gonna be more difficult. Well, I just, I was just gonna say, I don't, you know, I don't think this is good for anybody that trumped himself, and that's only a temporary flash, I guess. Um, he roiling the markets worldwide.

That's what he is gonna do. That uncertainty isn't good for the economy. It's not good for the tech companies.

It's not good for the country. And it's certainly not good for who you're talking about guy, the people who have to use this stuff. So, um, I just wonder at point, this breaks, what, when does this break?

When does it break? When does it enough? Or when does something happen?

There's like a, a critical mass and, and, and it just, it falls apart. I don't, I mean, am I being hopeful to think that that might happen at, at some point? Um, and, and there's a, a shakeup in our government or somebody.

I don't, I don't even know how you leash him, uh, pull him in from this kind of stuff. But he's, uh, there's just gonna be such long-term, um, impact on, on everything. I don't mean to sound.

So my next question is, is, is there gonna be a shakedown of Samsung to follow this? Because that seems to be the next thing. I mean, well, he doesn't consider Samsung an American company, right?

The concept of an American company versus those guys. We made t we made TMSC commit to building plans here in The US suspect. Well, because we hold the gun to their head that we won't help them when the, the, the Red China people invade, right?

I gotta bring this back to the, I gotta bring it back to the flap in the, the lips purpose of the statements is publicity trump's own brand image power. It's not an actual move of any kind, other than that. So, no, it's not Samsung, because, you know, I I don't think it's Samsung either, because to, to Samsung is not this burnish brand, this amazing brand like Apple.

So name a bunch of amazing brands, and yes, but Samsung. Samsung, However, Samsung is not, Samsung is not an American company. But there's one other aspect that we haven't touched on.

If Donald Trump was in the White House press room or here domestically, and this is a domestic thing, it would still get a lot of play. But, you know, he's here doing his thing. But no, he was actually at a press conference with the, the, the Amir of UAE, and he brought this whole thing up, the Amir's looking hi at him.

Like, what, what have you seen the video of this, the visuals? They're looking at him like, where, because he also at the same press conference brought up, what a great job he's done with the price of groceries. And he said to the Amir, I don't know if you know what groceries is, it's an old word or something like that.

This this buffoon right? Is out here in the world not realizing the jokes on him. They're not laughing with him.

They're laughing at him, but they know all they gotta do is throw a couple dollars his way. And like I said, he rolls over. And, you know, one thing that the, the oil rich Arab states are really good at, it's why they're still in power.

These Amirs and sheiks and all of them, they've always bought their way in. They've always bought their way in. Have we forgotten what happened to the reporter, Sean k Khashoggi kgi, right?

Yeah. You don't hear about that anymore. He's bought his way in.

They've done this. This is what the Amirs and the Sheik and the, the, the Saudis do when they find someone who's viable, they buy 'em. And that's what, and, and be proud America, you've been, your president's been bought.

Let's take a break. We'll come back, we'll talk about csa. Hey, folks, we're backing.

Well, the fun never ends when it comes to anything related to Washington. And there's been another turn in the CSA drama. It's a bit of a soap opera these days, but apparently there was a contract that got canceled that was, um, being disputed anyway, but now we're just basically saying, well, it's all mute.

Because anybody who was involved in actually needing that service from a company called Lidos, well, they're not working there anymore. So, Terry, what is going on in CSA from your perspective? And is this gonna be just, you know, continuing drama?

Sure. Uh, well, you know, more shenanigans basically from this administration when it comes to, to csa. I mean, I, I think really all of this stems back right to, uh, uh, when Chris Krebs said that the 2020 election, uh, was secure.

And, um, that sort of undercut, you know, Trump's narrative that, uh, he had, uh, really won the election and whatever, um, and Joe Biden had lost. So I think, you know, going back to that, we know that he's going after Krebs already for, um, you know, what, whatever it is. Um, I, I don't think there's any case there, but, um, he's publicly trying to skewer him.

So I think this is a bit more of the same. 4 billion contract was under dispute, um, already in the courts. But that's because Night Wing, uh, who was competitive for this bid, um, uh, and, and, and Lados, um, were sort of duking it in.

But, um, the Department of Homeland Security has made it clear, or it's, they've said that this has nothing to do with that. It's, uh, they're pulling the contract simply because the nature of CSA has changed, the personnel, the mission and, and whatever. Um, and it's a, it's a shame.

I mean, this, this contract is, uh, for agile cyber security technology, uh, technical solutions, right? So that's supports, uh, analytics testing, integration of security tools. Um, at, at this point, it's, it's gone away.

Uh, honestly, it feels to me, um, again, like it's, it's just this, uh, administration's grudge against cis a and also, um, kind of fits into what Chris Murphy, uh, I know you guys saw that, how he skewed Christie, uh, Nome last week, um, and said that basically DHS has turned into sort of the, the border security. I mean, that's where all their money is going. That's where all their effort is going.

I think they don't care about taking, uh, uh, money away from cybersecurity efforts at a time when we probably under heightened threat. So, look, you put a puppy killer in charge of these things, right? And this is what you're dealing with, again, the, the, the ludicrous nature of our lives these days in the us right?

You have a woman who's vastly, vastly underqualified here. Other than that, she sucks up to Donald Trump, and she shot a puppy when she didn't like it. You put her in charge of the DHS, and, and you make their primary mission to be border control, as you mentioned, Terry.

But it's not just the DH S'S primary mission. We're taking away troops in Europe and all over the world, because the primary mission of our armed forces are also to be border control. We, we have this is, this is like outta the, you know, I feel like an iron curtain has descended upon the continent, right?

That that's what you, this is where we are now. Csa, which has probably done more to help cybersecurity than any government agency or initiative that I've seen in my 30 years of doing this, has been bludgeoned bludgeoned. I mean, she, she, but here's the thing.

You gotta give these people credit. They're so transparent, they don't even hide it. She got up at RSA and said she was doing this and, and proud of it and proud of it.

People are sitting in the, the audience. I was there. They're sitting there shocked that this woman has the audacity to come to RSA conference where the security world gathers to tell us that they're, they're, they're disarming csa.

So what do you expect? What, what, you know, when someone has said this before, if it walks like a duck and quacks like a duck, it's a duck. If it sounds like a fascist and acts like a fascist, they're fascists.

That's what you're dealing with. So, I am I surprised by this? No, I do hope that private industry can fill the vacuum private things like R-S-A-C-R-S-A community and some other Linux foundations and, and some of these non-governmental entities, because there is a vacuum to be filled here.

'cause I fear we are going to have some sort of critical infrastructure or some sort of catastrophic cyber warfare incidents because we, we are taking down, we're, we're laying down our arms, we are laying down our arms and saying, as long as you're not brown and coming over the southern border, we don't care what you do. The part of this that's crazy though, the, the part of this that's exceptionally crazy is that there's a sort of this assumption coming from them that says, um, you know, those, uh, attacks from Russia or China that involve any kinda, uh, social engineering and propaganda are not attacks. Right?

Those are not within the purview of CSA to go defend. And that part is crazy because you're allowing foreign entities to pollute conversations and attack election systems with nobody there to defend. Yeah.

But if you, You know, it's almost like We a little bit open now. We're wide open. Everything's Backward, everything's backward.

We, we look the other way, yet we accuse CSO of being Ministry of Truth, and we denigrate Chris Cribs, who was great at his job. Well, we, we dismantled this organization. Sorry, sorry, guy for interrupting.

No, no, no. I was, say, it struck me, um, the, the, uh, which I think Alan alluded to this really straightforward nature of this. They didn't say flight.

They didn't go quite there and say, we don't need this 2 billion do dollar contract because we, the agency that would use it is basically destroyed. And, uh, it, and stood down. But they said the priorities have changed at cisa, which is Right.

That's actually right. They didn't even say, oh, we're gonna save a bunch of money. And one, one thing that was really interesting, uh, Terry, uh, you mentioned the, the, the Chris Murphy, uh, and the, just the general engagement that, um, that Christie no had, uh, at, um, uh, with Congress last week or two weeks ago, whenever it was, um, because he pointed out to her that they've spent all their money for the year, more or less already.

Yeah. They're, they're gonna be spending more than they have, and they're not to do that. So They'll spend more and then, and then they'll talk about what a terrible deficit the previous administration left them in.

Yeah. But that's a whole other thing that's going on here. They're not proposing budgets for next year.

They're spending in, you know, way like, what was it about a $500 million ad campaign that, uh, that, that they put up at, uh, at Homeland Security? I don't know how much of their budget that is. Probably not a whole lot, honestly.

And canceling a a multi-year contract might sound really great with all these billions of dollars and stuff. That's really not gonna make the whole, it's all theater. It's all theater.

I'm, I'm gonna start talking like Alan now. It's all theater. The guy they put in, in charge of fema, which I believe also is Homeland Security.

He, he came out the other day and said, look, he's never done this before. He really doesn't have a hurricane or emergency response plans put in place. They've lost a lot of people.

Give him some time. He's working on it. It's hurricane season here in two weeks.

Well, yeah. And as What, what catastrophe, what could go wrong? Um, I'm sorry, go ahead.

I was gonna say, as a native of Louisiana, that actually makes me shutter. Um, but, but I think, you know, to your point though, Mike, too, I mean, if you, if you admit that these things are going on that Russia's doing this, or China's doing that, or, you know, north Korean hackers or whatever, then, then you are admitting that they are our enemies and that they probably had something to do with, you know, election interference previously and all that. This administration can't admit to, to any of that.

So I think CEASE is gonna, whatever is left, the cease is gonna always be in the crosshairs. And I think we're not gonna, uh, you know, really admit publicly to these problems until we have some big incident. And to your point, you know, Alan, yes.

It's like an iron curtain dropping down, but what can penetrate that iron curtain? It's cyber attacks. Yes.

Yeah. Crazy. All right, let's take a break.

Can we talk about something not related to this administration? My blood pressure is, is high. You are watching Text on gang.

Let's come back and talk about Salesforce or something. Discover Textron Group, the epicenter of tech innovation. We are your go-to for reaching IT, leaders and practitioners worldwide.

Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients.

Let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Techron Group. Hey folks, we're back and we're talking about an acquisition that Salesforce made where they bought a company that specializes in AI agents.

It kind of took me by surprise. 'cause I could have sworn that I had just spent the last two years listening to Salesforce about how great they are and AI agents, and then they turned around and bought somebody who, well, looks like they're pretty good in AI agents. But John, am I being cynical?

Um, no, you're not being cynical. You're being brutally honest, Mike. Um, thi this seems to be a pattern, right?

We're, we're seeing the, all these AI related acquisitions to kind of fill gaps or add spackle to what companies have told us they can already do. So it kind of raises, its a little bit of suspicion. Um, so Salesforce basically says they've, they've agreed to acquire convergence, um, dot ai, which is behind these advanced systems that perform complex human-like tasks and digital environments, such as ai, agent design, autonomous tasks, execution, adaptive systems.

And in a sense, basically they're, they, the acquisition is meant to advance what agent force, the foundation of Salesforce's AI strategy is supposed to be doing, or has been doing, maybe struggling to do. It's interesting, I think Mike, a couple weeks ago you wrote a story. Uh, Salesforce has been just clobbering us over their head with announcements, but they, they, uh, they made this, uh, they, they had this continuing EGI initiative that they've been pushing, and, um, they, uh, made a couple of announcements including CRM Arena.

And, and I, it's, to me, it's interesting. I I think we're gonna see more of this. I mean, literally every other day we're seeing a minor acquisition, uh, by a large company that claimed to have a comprehensive AI agent strategy.

That's my takeaway. Um, so I too share your cynicism, right? It does feel like a rollup is already starting to happen in all these AI startups where the bigger companies are just buying them.

I guess maybe they figure they're less expensive now than later, or maybe the pricing is right. But, um, I don't know. Alan, do you think the incumbents have an advantage when it comes to agent AI because they have all the data, or, you know, is this gonna be, the startups are eventually gonna take over 'cause they're just faster and better innovative?

So, you know, last week, last Thursday, I did, one of my shimmy says my weekly shimmy says, uh, on LinkedIn, there might be a short of it on YouTube. I'm gonna get these all moved over to text on tv. I addressed this, right?

I addressed this issue, spoke about this one a little bit. Um, Mike, the startups never really take over. If you are lucky with every wave, one or two of the startups make their way into the old boys club, but it's still an old boys club, right?

And this is the way of it, right? These startups innovate the little fish, they innovate the medium fish. They, they buy that innovation and they make products out of it and product market fit.

And then the big fish eat the medium fish, and they put those products into their platform. That's what this is, right? And as a Yankee fan, Mike, I'm surprised that you don't see why Salesforce would buy conversion.

You could never have enough pitching or left-handed hitters in Yankee Stadium. And you, if you Salesforce, right? If you're Salesforce, you can never have enough ai, uh, agent AI expertise.

But Here's the Thing and why Yeah, go ahead. Go ahead. No, no.

Why do you need that stable? Why do you need that stable? Because, you know, the number of baseball scenarios is infinite.

And so you're just trying to cover as many as you can with different, even, you know, uh, uh, two lefthanded slider Pitchers can have different approaches that you might wanna use, but one may get hard. So you have an infinite bullpen you would like. Can I just wanna, I just wanna help us step back a little bit on this.

I spent, uh, uh, the last two weeks at two successive conferences that talked about AG agentic ai. And do you remember on this show, I've said before, there's no such thing as an AI application, not posted for rants about this before. There's applications and they use ai.

So Agentic AI is an agent that uses AI and one or more AI services, but it's an agent, no, CIO or CTO ever went around like five years ago saying, Hey, you know what we need? We need more agents. The question is, the question was for what?

For what? I need an ops agent. I need, you know, maybe, uh, a workflow agent, like there are lots of agents all up and down the stack, so to say agentic ai, it's just a really non-descriptive category.

Why did Salesforce buy an a, a agentic AI specialist? This is this human thing, right? This, this, they act as humans, right?

That gives me the willies, but let's just go with it. Um, well, I maybe ultimately the idea is to replace entire Salesforce is with Salesforce. I don't know.

Well, how about they did it? Because they can. Yeah.

So, so there's just, yeah, I'm gonna disagree in the sense that, you know, I'll take the opposite attack. What I don't need is 50 applications to do something. And, and, and every day I encounter something where I'm like, Jesus, I gotta log into this other thing over here just to pull that out.

So we connect something to this. And I think, you know, in the future, I think most people are just gonna wanna have an AI agent that pulls what they need regardless of what app it's in. And all these apps just become headless services.

You're never gonna have a one AI agent that does all you're look talking about a master agent. I will, I will have a master butler that will talk to all these other Agents, to the other agents and that, and that, and that is something, Salesforce and ServiceNow and the Automation Anywhere are all, that's the holy grail, right? This orchestrator.

But let me, let me take another tack on this. And again, I spoke about it on my shimmy says, we are definitely in some sort of m and a wonderland, right? We're seeing deals coming down the pike.

And every one of these deals have the word AI or a argentic ai, general ai, generative, a, they all have AI involved. And that's what's making it go. Now, is this a good thing or a bad thing?

Is it the sign of a healthy AI ecosystem in the natural order? Or is it the sign of a sick, there's some sickness here, or there's some bubble or what have you, right? Some malaise.

And you like that word, right, Mike? I do. And I'm reminded of Joe Kennedy getting out of the market.

'cause the shoe boy told them to buy stock. Exactly. That's exactly it.

It Wasn't, wasn't MA's what wasn't Malays Jimmy Carter though? Jimmy. Jimmy, yes.

That was Jimmy Carter. That was Jimmy Carter. That was Jimmy Carter word.

It was God bless Jimmy Carter. But here's the deal. The money being paid for these AI companies, whether they're really just an open source database company or something else, right?

Is is healthy? Are you paying a million dollars for a Postgres serverless? Postgres, a billion dollars.

Excuse me, billion. I don't know what the sale price was on. They didn't just, they didn't disclose it.

Right? Well, we'll find out. They're public.

Can I just mention, can I mention one thing? Sure. That it's kind of a pat on the back for the Futurum group.

Um, Salesforce was mentioning, again, the digital labor force and the digital, uh, labor market. And they cited the Futurum group, uh, report that came out that estimated it to be six. Wow.

$6 trillion by 2030 Seems pretty high. But, um, anyway, they're, they're using that as, as one of their, one of their, um, Yeah, but I, I think we're seeing that pull back because I, I think what's happening there is people are realizing, and, and you know what, speaking of future, actually Daniel Newman wrote this on LinkedIn the other day too. And I, I commented on it that at least between now and let's say 20 28, 20 29, what AI is gonna be much more of a copilot than a pilot.

It's not gonna necessarily take away people's jobs. It, it's gonna augment people and make them more effective. That's not to say that at some point in the future, it is gonna take people's jobs.

It will, but not, not in the near term, right? Not in the near term. It it's not, you know, for all those people running around saying it's gonna take everybody's jobs away, that's not happening right now.

But again, is it a bit of a gold rush and people are trying to get in before we find out that it's the man behind the black curtain pulling the levers that is making all this Wizard of Oz stuff, right? And then, you know, if you're not one of the first companies that got in on this gold rush, you get some pretty meager rations, right? You can start seeing fire sales and, and all of this stuff.

We, we've seen this cycle before. It's boom bust. And, um, right now we're certainly in the boom, the valuations are really high.

If you've got a good, you don't even need a good business. But if you could slap an AI angle on it, one of these, So what motivates this Alan or Chi? What, what, what, what motivates this?

Uh, I, I feel like this feels, well, o okay, right? But I, my sense, which is based on virtually nothing is Benioff is paling around with his, you know, VC, angel, whatever like that, the whole finance area. And he wants to be able to go and say, Hey, I bought this company.

They're really cool. They do all this stuff because it helps keep the money flowing. It's not marketing to, you know, the market customers, buyers, partners.

It's marketing to the financial c Silicon van value Community. But I dunno, I don't, you Gotta look at this as a business guy. Every company Benioff takes off.

The, the board is one less company on the board, one less potential company that one of his competitors are gonna buy or that maybe will strike Lightning in a bottle and become a, a real competitor to him. Right? He's taking pieces off the board when you are the king or the queen on the chess board, right?

And you have that advantage. You want to, you don't mind trading Rooks and Knights and Bishops because you know, you've got the power, right? As that board thins out, you, you have a strategic advantage.

And, and he does. 'cause he was early in it, he's got lots of resources. He's taking chips off the board.

I'm gonna be deeply disappointed when that $6 trillion in, in GDP or whatever it is, only turns out to be a measly two or 3 million. I mean, come on. Two or 3 million or a trillion.

Hey you, another 1, 1, 1 thing about Marc Benioff, you know, God bless him. He, I, I've known this guy forever. If there's ever somebody who is going to hammer home something in terms of marketing, in terms of mind share, in terms of trying to oversell an idea, it's gonna be him.

So that's why we're just gonna see an acceleration of a lot of AI related news. They also announced it on Thursday, last week, uh, Salesforce announced new pricing system around AI products. So more to come.

Yep. We, I, I, I think, you know, we're gonna see a deal a day here for a little bit, and then it'll be interesting to see, though, if the multiples stay high, right? What, what the, what the, you know, sell prices are.

Anyway, guy, I hear them calling you Gate. Maybe we should just have a segment that's deal, deal of the day segment. I, we, I think we, we have the last three shows.

Yeah, Let's make a deal. I'll give you $50 if you have a paper clip in your, Tell you what, next time I have the chance too, Alan, I'll, I'll, uh, I, I'll actually do a walking appearance and I'll get on the plane and sit down and That would be very cool. Yeah.

Sort of, well, I gotta Shade Moj Simpson. I run through the airport. Can I get an AI image of Marty Hall for this new segment?

We could probably come up with that. We come up with That. That's a good idea.

And Jay, what do we have for our Guest? That's a really good idea, Right? Um, I don't know how many people got that one.

I know Mike and John did. Um, all right, let's call a, a wrap on this version of, uh, Textron Gang. I, I'm, I'm, I'm afraid they revoked my, my, uh, my TSA pre or something.

Um, we will be back tomorrow hopefully with even more great commentary on what's going on in our world. Most, mostly about, uh, what's going on in tech in our world. But until then, is Alan Shimel for Textron Gang.

Have a great day, everyone. Hey, everyone. We're back here with what I think is gonna be our highlight of our coverage of the Imagine Conference here for Automation Anywhere in a beautiful Conrad Resort in Orlando.

I want to introduce you to Mihir Shukla. Meher is the CEO co-founder of Automation Anywhere. He's doing this 20 years, right?

For a lot of you working or watching this out there, you may not have even been working 20 years, let alone at one company, 20 years at my age. I, I can relate, but Meher welcome to Text Drug tv. Thanks for having Me.

It's my pleasure. So let's start right off with that 20 years ago. Yeah.

Talk about the vision 20 years ago. How's it, how's it changed? How's it stayed?

The same, how do you, when you get outta bed every morning, what get what still gets you excited? Yeah. That, that's a great place to start.

Uh, from, from, from day one, today to today, the mission has been same. What has changed is along the way, we had to invent few technologies. Some technologies got better in the industry overall, and as a result, we are able to achieve our mission better and better every day.

And that mission was to reimagine how work happens. Prior to Automation Anywhere, I had a chance to do four different multi-billion dollar journey. And by the time I had seen a huge part of the world, and I saw a part of the world where 70% of knowledge workers were sitting in a cubicles and doing work that you felt were in the human jobs, and they could do so much better if you unleash human potential.

And I grew up in a small town in India, and from my own experience, I knew that talent is evenly distributed. Opportunity is not. And the opportunity that was made available to me made many things possible for me.

So I didn't have to read a book to know that. Mm-hmm. It is my own life's experience, lived experience.

So the goal was to invent a set of technologies that take computers to the next level in how automation and AI can do things. And in doing so, the biggest bigger vision was could we reallocate intellectual capacity of the planet to more worthy causes? What if we don't have to process invoices and claims and few other things?

Could we be doing something more exciting, more fulfilling with our lives? And, uh, so with that objective, we started in 20 years, more than 20 years later, here we are. Uh, I, I get up every day.

And how often you get to real help, help relocate, intellectual proper Absolutely. Capacity of the planet. You know, I always say almost not multi-billion, unfortunately, but I done four or five, uh, venture startups founded, co-founded.

And I, I firmly believe that every founder in their heart Yeah. Believes that in some way what they're doing is making the world better. The vision you've enunciated here.

This, this isn't a small way, this isn't a big way. Yeah. Right.

Free us up from doing these repetitive, wanna call 'em low value. Yeah. Or they're not low value.

You know, we, we interviewed one of your customers from a light net, a a light, I believe they're here. Yeah. These are people who are processing people's benefits.

Yeah. Medical benefits, health claims. Yeah.

Mission critical life and death. That's right. They're doing so much with automation and, and, and automation Anywhere to speed that up, to allow people to go through higher value things.

And these claims can just get processed automatically. We've all been there. You submit a doctor's bill and you gotta wait 30, 45, 16 days for it to go through, not with automation.

Now, of course, this past year you've been in business as long as I have. This is sort of a, a high watermark, a a threshold year where we're starting to see Yeah. Maybe people's visions for what AI can be Yes.

Become real. Yes. This whole new thing.

Uh, a ag agentic process information. Yes. A PA Yes.

Talk about how that's fundamentally changed, R-P-A-B-P-A and all of that, and how it, frankly, it's fundamentally changed your company. Yeah. Um, so, so the in general, my view, uh, having done technologies for many years is that technologies take a huge step change.

So if you look at last many years, what we have been able to do is use, uh, RPA robotic process automation, document automation and, uh, task mining and various other capabilities as part of the automation platform. What all of these CAP capabilities were able to do combined is they were often able to automate 40% of the processes, but the other 60%, uh, were left, uh, uh, still manual. Suddenly, with the power of generative AI coming in, you were now able to take 40, what was 40%, sometimes all the way to a hundred percent or 80%.

So just a one new ingredient, completely chain the value proposition of what is now possible, not by itself, but in combination of everything else that existed before. And when you add this ingredient, it changes the game. To me, it's almost like if you, for people who are into cars, you know, they sell that STPI dunno if you're familiar, STP gas added of you added to your gasoline and it boosted the That's, That's right.

That's a good one. That to me is what we run here. You already had gasoline.

Yeah. But we just put in some extra octane that takes this thing off and, and it's an important, it's an important, it, it, it, we can't underemphasize how important that is. Right.

This is this game changing kind of stuff, but yet there's more Wait, there's more. Right. We, people are talking about artificial general intelligence.

Yes. And I'm not here saying it's gonna be next year or five years or whatever. I, and I don't play that yet.

Okay. Yeah. But you don't have to get that big a GI, if you will.

Yeah, yeah. To have an a GI sort of, uh, influence in what you are doing. Yeah.

Talk about that. The, the, I am I'm with you there. I'm not a big fan of word a GI by itself because it's an abstract concept, concept with, you have no idea what what that even means.

But I think if you define a general intelligence in context of specific purpose, so for example, if you're, uh, if you're, if you, if you're an intelligence for a self-driving car, the, the, it's, it's very, very clear that this car have enough intelligence to drive itself, yes or no. Right? So similarly, what we are focused on is that can we develop enough general intelligence to do, do vast amount of work as we call it, for a knowledge worker?

So can we give, uh, one of our, uh, one of our software, uh, uh, uh, mortgage applications to process and claims to process and supply chain and tax and audits and vendors, and a vast amount of these things that can, you just give it to the system and it has enough understanding of how to do this now in define that way it looks possible to achieve it. And we recently announced that We took a, for a step closer to it, towards it. And it is amazing to see how fast that is moving.

And, um, y you, you, you, you're never sure, but it, that, that day looks closer now than it looked a few years ago. It's, it's not, I I say the same thing about quantum. When I first started talking about quantum computing, I said, I'm not gonna be alive.

Yeah. By the time, but all of a sudden mm-hmm. I, I would not be surprised if Quantum's here in 2028 or 2029 even.

Yeah. I think it's the same thing. Same thing here.

Yeah. Same thing. It it all of a sudden that horizon's gotten a lot close.

Closer. That's correct. So, Meha, you know, we're sitting here at this conference, beautiful conference center.

You could feel an excitement when you go down by the, the, uh, stages and in the area. Yeah. For you, what are the big stories for Imagine this year?

What are, what's the big message? I think this year, as you said, was very pivotal for us, and we announced three huge, uh, uh, uh, announcement, uh, that moves the category significantly forward. The first one was an announcement to a, a significantly expanded, uh, agent tick process automation platform.

It's one unique element among about a hundred other features. That unique element is a process reasoning engine. Yes.

It's a very powerful capability in it. It, it, it works the way enterprises needed to work, which is in context of my enterprise, in context of my past information, in context of regulatory environment, reason what I should do at this point in time. And that process reasoning engine is what every enterprise's needs.

And it, it, it is the engine that will power all enterprise processes. And we announced the capability of how far this reasoning engine has come and how transformative it'll be for work. The second announcement we made was an availability of agent solutions.

Uh, these solutions are not like your typical solutions and applications that you see that earlier we have seen for, uh, 20, 30 years. We frankly don't like some of them because you have to often do 20 clicks to get something done, and it feel like they're designed for an era 20 years ago. I think there is an opportunity here to reimagine how, how this work gets done.

It can, can you have a solution or application that is agent first case, autonomous first, and design with a very different mindset. So today we announced four different solutions and many more to follow, including many would be offered by our partners. So, um, we are looking forward to how that transforms the industry and accelerates the journey.

Um, the third announcement we made is about a very unique set of capabilities in most cutting edge of the products and our services and partner services to create a offering called Autonomous Enterprise. Now, this is, this is designed not, not necessarily for every customer, but we have customers who come to us and say, I want to get to this vision of how, how future, how, how, what future of the work is. And I wanna get to a point where 70% of all my current work is either fully autonomous or assisted.

Give me a choice to get there in the fastest way possible. I don't want to take the long route. I want to, I, I wanna take a flight and get there.

And so we, we, we, we offered with our partners a uh, combining many capabilities and, uh, we take this to those customers to take the extra, the journey even faster. So those are the three announcements. Excellent.

Two more areas I wanted to touch on with you. One is, as, as, as we enter this, as of agent ai, you know, we've seen announcements from Salesforce and, and ServiceNow and, and even some of the big hyperscalers Yeah. Everyone recognizes that we're gonna be deploying multiple agents.

Yeah. Maybe dozens if not more. Sure.

This makes for chaos. Yeah. Right.

Yeah. We you need an orchestrator. You need, you need a, a commander, a master agent.

Yeah. Everyone has their own name for this. Yeah.

Now I know, uh, automation Anywhere recognizes this. You're also working with an orchestrator type of layer Yeah. Strategy.

Yeah. Would you expand on, without getting too far in the weeds, but would you expand on the need for the orchestrator? Yeah.

And why you think your company is, is in the, a good place to be the provider? Yeah. I think it is important.

You, you mentioned that there will be many, many agents, but in our view, there are three types of agents. Uh, the, the first type is personal productivity agents. And maybe many of the viewers understand that they're using one thing or the other.

Today, the second types of agents are, we call captive agents. They're specific to certain application platforms like Salesforce and ServiceNow or others. And there will be many others.

But the third type of agents you call the super agent or the commander agent, that is the trick. That is the power. That is where the power lies.

And Automation Anywhere is going to own that piece, uh, to orchestrate the, all the, all different types of agents across multiple application and making sure work gets done end to end. Uh, it is important for our, for our vision to, to, to, to, to realize is that we have to break silos. We have been working in the way since a post World War II era where everything is in siloed.

And that is, that is how, that is why the work is as, uh, ineffective and boring as it has been. Uh, there is an opportunity to change it with Automation Anywhere, orchestration, power, and some of our inherent capabilities. And, uh, we, we showcase that Imagine an ability to perform and orchestrate this work almost at a speed that you can give a command to and can work happen at that p that that pace.

And yes, it is possible for work to happen at that pace if you use automation. Excellent. And, and that the announcements from Imagine this available now or coming soon, or, Uh, every single thing we showcased, we, we was actual product.

We showcased it. Uh, many, most of them are generally available. A couple of our capabilities will be available, generally available soon.

But they were all available in beta and customers had a chance to play with it. And, uh, there was very gratifying to see as well. Another area I wanted to touch on with, you know, I speak to a lot of CEOs, big companies, startup companies, everybody wants to be a platform.

No one wants to be a product anymore. Products aren't good enough. All of a sudden I gotta be a platform.

Yeah. But you need products that work on a platform. Yeah.

Why is Automation Anywhere the platform, you know, you hear talking to as many customers, there are partners here and, and you know, people who are doing products that work in automation, the professional services that work with it. Talk to me about the vision for the platform. Yeah, I think I, what wanted do is make sure I mentioned this earlier, but we, we, we are, we, we are the best in class platform company and always have been.

But I mentioned earlier that we are, we have announced many solutions and many more to come. So on top of being the best in class platform company that it can use, we are now this new era of agent solutions that a business can use to, so make sure we are serving our customers IT and business both through two different, uh, offerings. Um, uh, if you, if you think about, uh, for for a minute, think three or five years out, because sometimes you can debate, uh, you know, future in three months or six, but it is probably all of us could agree what it looks like multi years out, multi years out.

The work is not going to happen the way it happens today. You are going to use, in my view, multiple LLM providers. Yes.

Because no one provider can provide you everything. It's perfect. Right.

You are going to orchestrate work across multiple applications. You are going to lead, go to the world that is as autonomous as possible and reduce reliance on manual processes. That's a goal we can all agree on.

Right? Agreed. You could de depend how much, but as far as you can go, you are going to do that.

Um, and, and, and, and, and, and you are going to need something that can execute mission critical processes. This is not about, uh, summarizing email. That that's, that's not what everybody, you, you process mortgages and claims and supply chain and you assemble planes with hundred thousand parts and mm-hmm.

These are important complex work. And you, you are going to need some help in doing this in a complex world with, uh, if you just mention this, if you just think about this four or five things and need to orchestrate this work, you're gonna need something for it. And that something is Automation Anywhere and it's available today.

Perfect. There you go. I love it.

Last question. 'cause I know you have to leave. I appreciate your time.

Yeah. Spend two days here talking Yeah. To customers, product, other partners, analysts press, boil it down.

What are you hearing from them? What, what's your take back from these people that you're gonna turn around and say, okay, this, this is where we may wanna double down. This is where we may have to change something.

What, you know, based upon the feedback, what, what have, what have you learned? I, I think a couple of things. One is that they are very excited about the products that they're using today.

And as a CEO of a product company, that's the first thing I want to hear. That we are delighting our customers and meeting their needs. Uh, when they heard our new vision and where we are going, they are super excited.

They can wait to try it. Uh, they, they can see the possibilities. And, you know, I've seen this every year where customers, uh, innovation capability is just unbelievable.

They, they can, they can use this product in a way that you have never imagined. And they make a difference in the world we live in, in a way that, that, that, that is so gratifying for us to see. So to, to, to, to interact and collaborate with these customers across 18 different industries, some of the largest companies, uh, on the planet.

And we can't wait to see what they will, will make of it. Uh, one of the key challenge is how to, uh, how, how, how to, how to help everybody understand the potential of it and how to, how to bring this to life in every department, in every function, in hundreds of different problems that the organization has to solve and how can AI transform it. And so our goal is to help together with us and our partners to help customers do that better every day.

Fantastic. Meher, thank you so much. We appreciate you taking the time to talk with our audience.

Continued success with Automation Anywhere. Can't wait for next year's imagine, but I have a feeling between now and then, this is gonna be a lot happening and going on, and it'll be interesting to look at this next year. Yeah.

In light of what is to come. Uh, thanks for taking the time to talk to us. Pleasure.

Uh, I'll mention one thing, Alan. The, in a technology space, uh, ever since I've started my career, the technology always takes longer than you expected and sometimes lot longer to achieve the vision that you set it Longer than the hype. That's correct.

Right. The hype is always out in front of it. What is unusual about the, what we are up to right now with Agent Tick process automation is, it is take, it is happening faster than we have imagined.

I've never seen anything week by week, week by week, day by day. And this is unusual even for us, for all. Imagine the customer.

Yeah. Oh, I, I I can't, can you imagine? That's right.

If we have a few more myth. Yeah. It's one of the biggest things, and I discussed this with the, with the a few people here today.

There's still the human element here, and the human element here is trust. Yes. If the human can't trust what this agent ai or general AI or any of 'em we're gonna do, they're not gonna let it do.

Yes. Right. And we're at this kind of awkward stage where humans are still learning to trust Yeah.

What the AI's doing. Yeah. And so we're almost pulling the reins back saying it's going too fast.

I don't trust it yet. Yes. But once we trust it, yes.

But the trust has to be earned. So one of the things we observed is we, we, we, we let our customers observe how it is doing it and put human in the loop where, where they, they have a chance to validate it. Mm-hmm.

Often what we observe is that after a person observes it for let's say three days, they get bored because this thing keeps doing the same thing again and again. And it's not as much fun. And Sometimes Boring's good.

Yeah. But so very soon. So if one person might, after watching three days say, you know what?

I trust this thing, let's move on. Somebody might say, I'll watch it for three months and then I'll trust everybody does it at their own pace. Right.

It is important for technology to earn that trust. Absolutely. That's a great way to end this.

Thank you. Thank you. Hey, I hope you've enjoyed our coverage for Imagine, uh, 2025 here in Orlando.

Many thanks to Automation Anywhere for having us here. We'll be continuing this story throughout the year. 'cause these are, this is, this is where it's at right now, right?

This is where the rubber meets the road with ai, agentic, AI and everything else. This is Alan Shimel. We're out.

Thank you. Hey everyone, it's Alan Shimel from Techstrong. We're here today at the beautiful Conrad Resort in Orlando, Florida for the Automation Anywhere Imagine Conference.

It's been an amazing two days. You know, there's so much going on in this area. You may or may not know Automation Anywhere.

They've been a leader in, in, uh, B-P-A-R-P-A for years and years and years. Um, but the world's changing and, and, and so is Automation Anywhere. And, uh, we're gonna talk about that.

I'd like to introduce you to Audi Ganti, right. Chief Product Officer, CPO at Automation Anywhere. Audi, thank you so much for joining us today, and it's great to have you here.

Thanks for having Me, Alan. So I mentioned your chief product Officer. Mm-hmm.

But I always like to give at the audience a a sense of your journey. Yes. Right.

How did you come to be the Chief Product Officer here? So, Alan had joined Automation Anywhere about little under four years ago, and I joined as Chief Product Officer. So as part of that product, technology community, um, those are my areas of responsibility beyond, uh, everything else that goes in a, you know, in a startup in a company our size that's growing so fast.

Um, so I've been here about three and a half odd years before that, I was at Salesforce, 15 years, um, joined back when it was one big cloud or one product line. Uh, SFA, uh, I've seen it through this various growth, uh, transitions and spikes and, uh, kinda the major growth areas. So join when it was, you know, sub 500 million left in 2021 when it was around 27, 20 8 billion.

So saw that growth curve always in products. Um, was in different product lines. Really at the core.

I love building, launching, growing product lines. Like that's my DNA. Um, and that was, uh, you know, Ryan ran 1,000,000,005 business, um, at the end of my tenure at Salesforce and decided I wanted to be in the automation market.

That's why I joined the leader, which is where I Am Mission Anywhere. So Audi, you know, to me that's a really interesting, uh, history coming from Salesforce to Automation Anywhere. 'cause these are two companies that are kind of leading the, the path forward in terms of agentic, agentic ai and using agent Salesforce has of course made a huge pivot That's right.

To, to this agent based thing. I, I have to ask you, four years ago, you're looking at coming to Automation Anywhere for automation, and of course automation is, you know, in the tech world, is always considered a good thing, right? Yeah.

Um, did you have any inkling at that point that, uh, the agentic piece of this would be such a, I mean, such a, a, a watershed event? No. To be, uh, very indeed.

Right. Uh, four years ago, no. Uh, but I always knew the automation market was right.

Really ripe for disruption. Mm-hmm. Um, yes, we are, you know, a market maker for RPA robotic Process automation, and there are other sub-markets out there, but there's no, like the, the, the technology had evolved, but with AI and the kind of the broader, uh, market moves that's happening across various sections, I did feel that automation, that's one of the reasons I joined automation anyway, was the market was gonna explode.

Yeah. And this was, something was clear to me even in 2021 when I was, you know, looking at, I spoke to Mihir and team about joining Automation Anywhere. Uh, because all of, you know, every business needs automation.

Absolutely. Um, as much as you might say that, you know, operational efficiency is, you know, especially now actually operational efficiency is even more important, but even more so growth. How do you drive growth by investing in the business?

And that's where automation comes in. Um, we are always been in ai, um, frankly, even RPA and kind of BPA, there's always been a lot of ai, more AI ml, um, but definitely this entire market of agents and gene AI before that is completely disrupted. Absolutely.

To me, I think of it is, you know, they were selling automobiles before Henry Ford mm-hmm. But the whole concept of the assembly line and the little automation, and, and so what, you know, it made a very different Yes. For a very different car market automobile market.

Now, we've talked around the edges about Automation Anywhere and, and you know, their history, they were a leader in RPA, they've been a leader in r they're still a leader and still a leader. Leader in RPA. Exactly.

Yeah. Um, but when, when, as you, as Chief Product Officer, when did you, you know, go to your peers at the exec team and say, well, have you gone to your peers, I guess is a better question and say, Hey, we, we have got to go all in on egen, or maybe not go all in on egen. Yeah.

So, um, we've had, so there's a timeline to this. Mm-hmm. We were the first into Gen ai.

Uh, in fact, we started working with OpenAI and some other companies before chat, GPT, and it's primarily to drive more automation use cases around doc, specifically around document processing. Mm-hmm. Um, because we found a way where we could, for example, um, process unstructured documents.

So this is like back in 2022 before all the craze of chat GPT, and Sure. You know, the, the explosion. Uh, and for the next six to eight months, and this is post chat GPT, we investors started investing more and more in that area because we saw amazing results.

Uh, and now our, one of our fastest growing products is document automation. You know, three x growth, uh, 65% of our documents are processed using generative ai. You know, it's, wow.

And we are talking about single digits last year and now 65%. So just crazy kind of growth there. But really last 18 months.

So we were first of all on Gene ai. Then we started thinking about, okay, how do we tune this or focus this automation outcomes? Because most of the gen AI you see out there a general purpose, you know, summarize an email or, you know, do something else.

Write a blog article. Write a blog article. For us, our customers were asking, well, how does this apply to me in my automation, uh, use cases, whether it is accounts payable or what it might, whatever it might be.

So our focus is, okay, what are the investments we should be making to use Gen AI and then AI agents, but tune it and focus it on process automation? And so that was kind of something we went all in, I'd say about 18 months ago. Um, we as a product and technology team, we've been working really closely with not only the hyperscalers, we AWS and Microsoft gcps of the world, but also startups who've been coming up and up and coming, like AI native companies.

So we've been kind of ahead of the market in that sense and getting access to early technology in before it's available in the market. Uh, so long story short, I would say is we, we kind of very much all in into what we call agent process automation, which is kind of our version of agent ai. Uh, and Mihir and the broad exec team, frankly, the company is all into, um, it wasn't overnight, but, but I think the last 18 months we've shown we are first to market with a PA first to market with agents.

We launched it nearly a year ago before agents are cool now, it's obviously everybody's talking about it. Uh, and obviously we'll talk more about some of the new announcements, but I think the first more advantage has really helped us. Absolutely.

Well, yeah. You were starting from a position of strength. Yes.

Right. I, I just wanna make clear, a PA is a gen process automation, and that's, that's kind of the new term here that we're using for this. I will tell you, I'm not as an expert on automation, obviously, as you are, but I've always looked, especially at RPA and said, it's great, but I always felt there was another shooter drop.

You know what I mean? That something like an AI agent would really just ignite this. Um, you know, I, I remember my first kind of exposure to RPA, I met someone in a startup group.

I think I was a judge at a pitch event or something. Yeah. And, and they had an RPA document mm-hmm.

Basically system that was really cool. But it, I just felt it needed the agent to really make it more independent, more, uh, yeah. Mm-hmm.

And, and now when you look at this, it, it, I mean, certainly it's, it, it's just as you said, when you go from single digits to 65% of one year Yeah, that's right. With the top line growing at three x, so it's, everything's growing. It's not the same.

So three x and I mean, you don't see that, right? This is like, when cloud came out, we didn't see this kind of growth. This is the kind of growth we saw when maybe the internet first point commercial, right?

Yeah. We truly believe this, the phase phase we're in, and I think this is not only in automation, but broadly in the market. This entire agent tech, um, gen ai, uh, push is like cloud or even probably like the internet.

I think it's bigger than cloud. It's more like the Internet bigger. Yes.

Uh, and so it's gonna make come complete not only disrupt markets, but you're gonna have the next Googles and Facebooks and Absolutely AWS is the world from this coming out 10 years from now. And so we want to be very focused on this. I agree with, you know, I, I try to tell people that I've been in technology for 35 years, a long time, and I've seen a lot of stuff come and go and come again.

Um, there are some things that as a technology person, get me excited the cloud. Mm-hmm. Right?

But explaining to my mother-in-law what the cloud is, all she knows is her pictures were stored up there. Right? Yeah.

Then there are some things that just change civilization, the internet, the cell phone. Right. These things.

It's not just my geeky friends. Yeah. Right.

It's, and, and that's to me what, what we're dealing with here with something like, you know, agen, a PA and, and agents. So, howdy. If you don't mind, I'd like to maybe pivot a little bit now and, and talk about the Imagine Conference here.

It's been two days chock full of product announcements, partner success stories, customer success stories, learning. Mm-hmm. You know, I think people are still, uh, wrapping their h their hands, their arms around a agent process automation.

What it, what it's more than just a name change. Yeah. Right.

What, what really does it entail? If you wouldn't mind, for all of our audience out here who weren't able to come Yeah. Who weren't here.

Tell us what we missed. Give them a recap, let make sure we, we, you know, let them know what, what's, what happened here. Yeah.

So it's been an amazing, what, two and a half days. We still got, uh, the rest of today. Um, we, I, I kind of bigger up three, three main things.

First, we talked about the amazing momentum we're seeing with Agent Tick process automation. This is something we launched last year as a category. It's really a new category, an expansive category.

Um, and we had R Donnelley, uh, we had KPMG. Um, we had Washington Post some major customers talking about how they're using agent process automation and really driving incredible operational efficiency, but organizational change and business change within their, you know, large enterprises. Uh, and across, you know, manufacturing and I talking about, um, uh, uh, different industries essentially.

So that's one big thing. And, uh, across, uh, even the sessions, you, we had a lot of customers talking about, you know, the business benefits and the gains they're seeing with HG Cross automat. The second big announcement was set of new innovations, and that, as the Chief Product officer I'm really excited about, first was around the process reasoning engine.

And essentially the way to think about the process reasoning Engine, it's kind of a secret sauce. It's, um, our unique IP that helps us drive differentiated process automation outcomes in the world of agents, because you have general purpose agents. Our focus is on driving process automation outcomes.

And the way we do that is, you know, we have our own unique models built, are tuned on, you know, hundreds of millions of automation runs on our platform. We have customer specific context, which name is personalization for specific customers, because, you know, customer manufacturing is gonna be very different customer financial services. It needs to be tuned towards them.

Uh, and then these more, uh, goal-oriented self-reflective agents, kind of cognitive AI agents that can, for example, um, pro, you know, look at unstructured content like a product catalog and make decisions on what's the right product, uh, replacement for a specific customer contract. So there are different use cases, obviously. Uh, and so that entire process, reasoning engine ability to then, um, access in, in a very expansive enterprise tool set.

So we're an automation company. Customers use us to build RPA bots, APIs, process documents, build AI agents, bunch of different things. We also announced, so that pre was a big part of it.

And, and underlying pre is also having an open ecosystem, open platforms. We announced partnerships with the AWS, uh, they were on stage as was Google Cloud. Uh, part where, where, um, we know part partnering with Google Cloud, for example, on the A two A protocol, uh, we've been partnering with philanthropic, uh, model context protocol.

So really having, kind of really focusing as an automation vendor on an, an open platform and open ecosystem, that's critical. So that's all part of the process reasoning engine. We are the first and the only automation, uh, vendor to have that.

And we believe that's critical for any company to drive differentiated process auto outcomes. The second big announcement was general availability of our agentic orchestration engine. So we truly believe one of the core beliefs around a PA is it's a combination of deterministic and, and cognitive.

If cognitive is driven by the process reasoning engine agent take, our agent orchestration engine is a major, uh, investment in our deterministic processes. The ability to basically run long running mission critical processes with the right governance and operational, uh, visibility. So that's gone ga um, with, with this summer.

And the third big announcement we made is, again, in partnership with AWS, uh, around human agent collaboration. So one of the things that in this world of agents was completely changing is how you and I engage with agents. It's not gonna be the same forms and traditional user interfaces.

It's not gonna be those apps, you know, where, you know, like a, frankly in a Salesforce or a ServiceNow, any of those, it's gonna be these more collaboration experiences, conversational experiences, but where you get work done, it's not a chit-chatting experience. Right. Um, and so how do you translate user intent into process action, but again, personalized to that user, personalized to the employee, and personalized to the customer.

And that's what we announced with our partnership with Amazon Queue. Um, which is kind of under the umbrella of human. Yeah, no, It's something we've covered a lot.

That's right. You know, it's funny you mentioned Salesforce in ServiceNow. One of the things we've seen is both of those companies, they don't want to just give you their agents.

They want to be their agent orchestrator. Yeah. You are the fir this is the first time I've heard someone refer to it as an agent orchestration system.

But in my mind, that's always what was needed, sort of the Kubernetes is containers. You need an orchestrator for, for your agents. And, but they may not call it that, but that's really where the race is, right?

Everybody wants to, because we're gonna have a multitude of agents. We'll have an agent that interacts between you and I, between a process and I, between, you know, we're all gonna have a, a fleet of agents, whether they're ephemeral agents that are kind of single use and disappear. And I spend another one up next time, or persistent agent or however terminology you want to use, these agents are gonna need to be managed.

Yep. Do you see that as a key part of the Automation Anywhere value? Yes.

Is managing this, this whole, this Agent Army, if you will. Yeah. The, the short answer is a definite yes.

Mm-hmm. That's along with the process reasoning engine, which is kind of the secret sauce for all agentic behavior. Mm-hmm.

The orchestration of these agents is critical. And we expect customers to build agents on our platform, on AWS, on GCP, on a Microsoft Azure, on Salesforce, on ServiceNow, on open source platforms and anything, because frankly, you're gonna build agents for different purposes or different different platforms. And that's good.

That's great. Uh, but how do you orchestrate those agents and how do you also orchestrate both, uh, agents and deterministic workloads? Because when you're extracting a document, you don't necessarily need an agent or you're calling SAP, you don't need an agent for that.

You can just use an API to call. It's Really, well, there's this no man's, that's a funny word to use for it. No man's land, but there's this gray area.

Yeah. Let's say better between what's an API call. Yeah.

And what's the n agentic? And you'll find a lot of, um, um, a little bit confusion in in the market. Yeah.

There is where everything is being called an agent. Uh, the way I see it is a cognitive task is an agent. An agent can call a set of tools like an API, an RPA bot and other things to do.

Its work. But that's ultimate goal is you want to do something which is probabilistic. There are many things that you don't need are probabilistic.

It's just, just, you know, in, you wanna call, uh, a Salesforce api, an S-A-P-A-P-I or a legacy system, you need to, you know, uh, use RPA bot to access it. You don't need an agent for that. So, but what we are focusing on, how do you orchestrate a process?

And a process is gonna have deterministic steps, it's gonna have cognitive steps, and that's great. And you can have some of those cognitive steps like agents built on our platform, some built on other platforms. And we wanna own that orchestration layer.

And like you said, the management layer. So how do you drive real time observability on how agents are operating? Are they actually responding as you expect them to respond?

'cause again, they're probabilistic. So even 95% success rate may not be good enough. It has to be 99 and more for certain mission critical processes.

So having the orchestration layer and the observability layer is all part of our agent orchestration system. I love it. And, and I You're a hundred percent right.

Make no mistake, the company that was orchestrating and managing these agents is in the catbird seat. Right. In the powered seat to, as this whole thing starts tolo unfold.

We kind of stopped though, but let's go back to imagine. So that was two big Announcements. So those are the two big announcements.

And the third I would, uh, say is some major announcements with AWS. I already talked about some of the product announcements. Yes.

With Q with with q uh, also with Bedrock and some of the other, uh, investments there. But also with Google Cloud around, uh, partnerships on various products, but also on the agent to Asian protocol. Uh, I won't say March, but in Google's IO event, they're gonna announce some joint, uh, investments actually happening, uh, next week.

So Excellent. They'll be announcements there. So it's customer innovation and some key strategic partnerships.

I love it. And, and it's interesting, you, you've got Google, Microsoft, and AWS integrations here. Yes.

As one should. Yeah. If you wanna play in that cloud space.

Right. Um, Howie, let me ask you to look in your crystal ball now. So this look, this has turned the, the RPA industry on its head.

Mm-hmm. Two years from now. Two years, 12 months to 24 months.

How does this manifest itself? So one, the one caveat I would put in is, um, this entire agentic AI market is moving so fast. I mean, two years ago, no, there was no concept agents.

No. Now we only talk about agents. Um, so I would put that caveat with this market every three to four, it's just turning on its head.

Right. Um, and I'm gonna put a crystal ball, like I gotta kind of guarantee that crystal ball, huh. But what I will say is I do expect, um, we've talked about some of the key, um, elements here, right?

Agentic orchestration agents being the probabilistic agents being, um, um, you know, you can orchestrate but also control plethora of agents across different platforms. I do see there, there's gonna be a big focus on how do you stitch together these various agents for specific solutions? How do you deliver value to the customer?

Because right now it's still very much a technology doc. Mm-hmm. Versus okay, what are the at scale deployments that can be out there?

We already had some customers at the RDS and the of the world talk about how they're deploying it at scale. I think we need to go, today we have about 1500 live deployments. I expect that to go into the many, many thousands, if not 10 thousands.

And that's what I think the next two years, the the maturity curve Yeah. Is gonna be important because ultimately, at least what we are in is how do we use agents for mission critical operations? And for agents to be used in mission critical operations, it needs to mature.

Yeah. It Has to be mission critical. Yes.

Yes. It has to, uh, it can, they can't be higher rates and there needs to be the right observability of the pieces. So our focus, and I think where the market is also gonna focus on a, across like our partners, the A W S's gcps of the world is how do we together partner to ensure our biggest cu all our customers can, um, roll out and kind of scale out these agent tech solutions and workflows at scale across all their business operations, and see the impact on revenue, on operational efficiency, um, as well as on lower risk.

Because those are the three pieces, right? Revenue, cost, risk, and there's three elements. We gotta see the impact of agents.

And that's where I think the maturity curve is gonna happen for the next two, next two years. But while we, while that happens, I think there's gonna be a lot of innovation going on in the market. We are just getting started on things like model context, protocol and agent to agent, uh, kind of, uh, protocols.

Like how do different agents interact with each other? 'cause you, we all talked about how different vendors are gonna have different agents, so how do they talk to each other? That technology we just getting started.

So I think there's gonna be a huge surge on innovation and commute, surge on innovation. But I think what the market will demand is, well, how do I scale this out with my company and actually show business benefit? And that's gonna be our focus.

Excellent. Heidi, we're about out of time. I wish we could talk more, but we will because what, look, my feeling is we're at the beginning of the beginning of the Yes story, not even the end of the beginning.

Mm-hmm. So I look forward to continuing this conversation in the weeks and months ahead. But congratulations on a fantastic imagine event here.

Thanks for coming on Techstrong and, uh, tech Strong TV and continued success. Alan, Thanks for having Me. Appreciate it.

My pleasure. All right. We will have more imagine coverage here from the Ooma Automation Anywhere conference in Orlando.

Stay tuned. Hello and welcome to the latest edition of the Techstrong AI video series. I'm your host, Mike Bazaar.

Today we're with talking with Sam Gong, who's Vice President Marketing for Work Span. And we're talking about this whole issue that emerges. It's called the Partnering Tax.

We all have experienced it. We don't necessarily call it that, but once we get into it, you'll know exactly what we mean. Hey, Sam, welcome to show.

Thanks Mike. Happy to be here. So, partnering tax, we see this especially among big tech companies, but I think we see it everywhere.

It's when a couple of companies get together two or more, and they kind of agree to work on something with a customer. It involves them meeting each other. And all too often from the customer's perspective, it feels like all these folks are meeting it the first time when they come to their office.

And it's not exactly a great outcome. 'cause nobody knows what the left hand or right hand, or any other hand is doing. So Sam, how will AI help us get around this or avoid this situation?

Because it does lead to a lot of lost productivity, for sure. Yeah. Thanks Mike.

And I think, uh, ai, AI is a part of the solution. But if we think about how we got here for a second, this is, uh, this is a problem that I think has been coming for a long time. And if you go back all the way to the beginning, uh, we used to have verticalized businesses, right?

You'd own the supply chain, you'd own the resources, you'd own manufacturing. The goal for the business magnets of the last century was let's put it all in one stack and own all of it. And then we can control quality for the customer, and we can control margins and prices.

And I think, uh, the trend in our century has been for problems to get more complicated, for technology, to get more complicated. And especially in the last 15 years. And now with ai, not only as part of how we work together, but as what we're offering to our, our customers, the pace of innovation has accelerated to the point where there is no one company in the world, not not Amazon or Google or Facebook or Meta or any of these companies that can, can own innovation at every layer of the stack.

And, and so what that necessitates is, if I want to provide the best solution to my customer, I need partners to come in there with me. I might need Nvidia for hardware and, and the the chip set that's gonna run the foundational l uh, LLMs. I might need Anthropic or Amazon Bedrock or Google Gemini, right?

Like, you've gotta piece together a solution that's best in breed for your customer, and that means that you're gonna work with partners. And so now when you present that, that solution to customers, uh, the complexity tax comes in because your sales team, your marketing team, everybody that wants to position and describe that solution to customers, they're not only trying to keep up with what you're doing as a business and what you're building, they're trying to keep up with how is that current in the market? How does that leverage our partner's strengths?

How have we put all these things together? And, uh, it just gets stuck, uh, in a single person's brain. There's too much input to, to stay on top of all of it.

So that's what we mean with the, with the complexity tax. And, uh, of course this is, um, this is a really hard problem to solve. You're not gonna fix this with a SaaS application with if this, then, then that logic.

Uh, you need cognitive tools. You need the, the benefits of, well, I can read a, a customer briefing from, from this partner and know how to digest that and boil that down so this salesperson can say the right thing at the right time on that deal. So that's where we see the intersection of, of AI and the acceleration of innovation, creating this, this pressure to go to market with partners, whatever layer of the, the stack, whether you're a service provider or a software provider or an infrastructure provider.

Uh, and then if you're an individual in that role showing up in that, that room, like you said, uh, when partners come together, you don't want it to feel like this is happening for the first time. You want it to feel like there really is a joint value prop and a joint solution. And we see AI having a big role to play in helping those partners collaborate as a, a unified front Your point, um, can I reduce the number of salespeople required for that meaning in the first place?

'cause I think that's also part of the problem. But if one salesperson might be able to, uh, accurately represent the others and then the others can go do something else more interesting. Exactly.

And I, and I don't think it's about reducing the number of people. I think it's about maximizing the leverage for people, right? And so all the, the customers that work span talks to that are running sales orgs and partner orgs, uh, you want to provide the best experience you can to every buyer.

You want to provide the most support you can to every partner. You never wanna say, oh, well that account's too little for us to, to take a look at. Right?

Or that partner's too little for us to take a look at. And so what AI helps with is you can take your people that have that subject matter expertise and they can divide and conquer, right? We wanna put the individuals in the driver's seat and say, well, I can put 20% of my time into all these accounts if AI can help answer some of the basic questions, and I can focus on these big deals or these big accounts.

So it's, it's not about reduction as much as it is about leverage and being able to spread every person in their role to cover more deals and more accounts and more partners. And we'll also work. Conversely, I assume that when there is an issue, it'll be easier to sort out, well, who might be the most likely one of this partnership who might be the issue or the root cause of set issue.

And, uh, we can kinda cut down that whole customer service conversation. And, uh, exactly. This isn't just a pre-sales problem.

Pre-sales of course gets a lot of attention. Everyone wants to increase revenue and, and money coming in the door. This is a full customer lifecycle problem, right?

And so if you succeed in selling that joint solution, you have to have a plan to provide joint support for that solution. And that that doesn't just make the salesperson's job more complicated. It makes the support team and the success team and the delivery team's job more complicated.

So we, we say it's a complexity tax and we start with the initial sale, but it's the full customer lifecycle. Every single person that's involved in delivering that joint value to the customer needs this support in their role to be able to pull in those partner solutions and and deliver the joint success in today's market. Will it become easier to onboard people onto these teams?

'cause I think one of the issues that a lot of organizations encounter is when they lose somebody from these teams, it, it's a, it's heart wrenching to them because it takes six months or more to replace them. Even if I, once I find the person, I, I think both things are true, right? When you increase the leverage for each person in their role, that's a good thing.

But it also means that potentially that person leaving means a lot of the knowledge walks out the door with them. And, and so what we're finding is, especially with partner managers, right? You think about, I'm sitting at the intersection of all the complexity of my company and my company's products and structure and how we go to market, and then trying to unlock my partner's, uh, solution and, and the value they can bring, but understanding their complexity and product and go to market and all of that.

You are, you are highly leveraged at that intersection. And it has not classically been a well documented process. It's a lot of emails, it's a lot of, uh, PDFs and, and agreements that get shared back and forth in, in spreadsheets and PowerPoint.

And uh, there, you, you can have a partner relationship management portal. You can have places where some of this stuff lives. But that true understanding of how are we driving this partnership very much ends up in, in people's heads.

And we're, we're, we're still early, but I think what you're saying will come true, where as that person who sits at that intersection works more closely with ai, then that person will have flexibility. Not if they want to change careers and go do something else or change companies, but also if they want to build another partnership and shift where they, they focus their time having AI act as that repository and they're working buddy buddy with, uh, with an a i teammate that they're training that not only increases their leverage, but it increases their, their flexibility to spend their time on another partnership and give the business more ability to bring it in backfills or staff that role with less of their time if they decide to do something different. Do you think as part of this, that the AI is essentially gonna function a little bit like the institutional memory?

Because we all have people in the company who've been around for a long time and they serve that role, but um, you know, when they go, so goes the knowledge. Yeah. I think, uh, institutional memory's a great way to put it, right?

And, and if you look at any process today, I, I work in marketing and even in the, I haven't even been at work span two years, right? But in my two years, I'm still uncovering institutional knowledge that wasn't part of my formal onboarding. It's not easily accessible to me.

It's three clicks down in some folder hierarchy that I just haven't explored yet. And we go to launch some program and I ask the team, have we done something like this before? And it takes a lot of digging to surface institutional knowledge.

And, and so I think one of the things AI will do across functions, not just with partnering, is help people and businesses understand each other faster and, and shorten the onboarding and, uh, make it so you can provide value with more context sooner. Because there's an AI that can answer your questions and do all that digging for you. How do I get started with this?

I mean, do I just kinda like deploy some sort of AI agent and then it kinda monitors this until I get that level of institutional memory? Or when is the length of time to get value in this whole approach? Uh, so with, with work spend ai, we, we launched this last quarter and we've, we've been taking our first customers through onboarding.

Now, uh, it really comes down to how accessible is your knowledge. And if you have all of the questions and answers, and you've built really good field enablement programs, you can feed those field enablement, uh, uh, call recordings, training decks. The AI will pick those things up and then very quickly be able to disseminate that knowledge to the field.

If your data's in good shape and you can feed it quickly, you get value very, very quickly. It doesn't take a long time of passive listening to be able to provide value. And if you haven't had that, then you need to do some digging.

You need to go back through your Slack threads and your emails, find the, the questions and answers so the AI can, can provide that answers that you would provide. You need to do more cleanup if you haven't already organized that data. But then I think, um, and again, I'm speaking with my own experience using AI and marketing as well as our, our partners using AI in, uh, in their roles.

You get into the habit very quickly where you say, oh, if I stay organized, I get this immediate benefit. And I'm not an organized person by default. You know, I, I have a million sticky notes here on my desk.

It takes me a lot of effort to stay organized. And one thing that's really helped me is when I see that immediate return on here's what happens. If I can provide all these inputs to the AI, that then helps my team, the AI trains me to, to provide better inputs and stay organized because the benefits are so immediate.

And so we've gotten great feedback from, uh, from our first partner managers that are training their AI and seeing that immediate, oh wow, I can, I can turn all of these questions that I used to have to go and answer on Slack over to my AI teammate now. 'cause I'm confident it, it'll say what I would've said in that situation and take those base level questions from my sales team and then I can go provide better support on the big swing deals that need to close this quarter. I think every salesperson probably has this experience where they have a call with a customer and then they hang up and about a few hours later they start kicking themselves and say, I should have said this, or I should have said that, or I shouldn't mentioned this.

Will they AI agent essentially remind them of things that they should say as the conversation's going along? Yeah, I think, um, across, across sales, we're seeing this compression where, as a seller, you wanna provide as much value in every touchpoint, right? You're an ambassador not just for your product, but for what it's like to work with your company.

And so tools like Zoom and Gong and all these call recorders, I think we're seeing what used to be a very long cycle of let's go back and let's do our Qs and look at our deals that we've won and lost and kind of do some, some introspection on why that happened. All of that is being surfaced into a much more real time immediate feedback. Let's dissect this call and have the AI look at your talk time.

io surfaces for every sales manager to coach their team on. Are you doing too much talking or are you opening up the, the space for your prospects to tell you what, what they need? And so, uh, with partnerships, same as with the rest of the sales cycle, that that kicking myself, uh, did I say the right thing?

Uh, I think all of that coaching now with AI and workman's AI teammates is gonna be at the level where you're not just running better deal cycles, you're learning to run better deal cycles with a much faster feedback loop. So I think there's, there's two orders of acceleration here, right? It's not just the availability of that information in the deal, it's also the availability of those best practices for sales managers to coach the team on, Hey, we could win more deals with our partners if we applied what worked over here on these other, other opportunities that we have in the pipeline.

All right, folks, well, hey, sales has always been a game of confidence. And if you have an AI agent that's helping you figure out what your partners are doing and what your customer needs and what's going on in the industry, you're gonna be a lot more confident. Sam, thanks for being on the show.

Thank you so much, Mike. ai video series. You can find this and others on our website, and we invite you to check them all out.

Until then, we'll see you Next time. Hey, everyone, we're back here at the Imagine Conference from Automation Anywhere, uh, in Orlando at the Conrad Resort. It's, it's a beautiful resort and it's been a, a fantastic two days of learning here at Imagine.

Our next guest is Micah Smith. Mic is VP of Developer Relations and Community. Nailed it.

That's it. You know, as I get older, it's harder for me to remember this. That's, That's one question down, One question down.

All right. They're not going to bing me on that one. Anyway, Micah, welcome to Techstrong tv.

It's a pleasure to have you on here. Um, we're gonna talk about a lot of things, but let's start off talking about you. Yeah.

So I invested that. Me too. Um, imagine, so Micah, give people a sense, I gave them your title, but you know, titles are fungible today.

That's Right. Um, Give them a sense of who you are and what you Do. Yeah, so I lead the developer relations and Community teams and Automation Anywhere.

Mm-hmm. For us, that means everything from our training, our Automation Anywhere University to our Pathfinder community. Uh, that includes things like our Pathfinder framework, our a Agentic Quest bot games, which are challenges that we create for developers.

Um, and basically a lot of community programs that we're running. So we just did a training camp for, uh, everyone to learn agentic process automation. We did 46,000 course completions in a total of weeks.

Wow. It was a huge accomplishment. Way more than we even expected, if I'm totally honest.

Um, so I do that. And our mission is really to empower automation programs to be successful. And that means empowering automation program leaders to know how to build and scale an automation program.

And it also means training developers to learn how to develop, deliver, and deploy successful automations. Sure. Let's unpack this a little bit.

So you mentioned, I think it was called Automation Anywhere University. Yeah. For those of the, our audience not familiar, give them a sense.

What, what is the university about? Yeah. So this is all of our efforts to train developers to learn how to build, deliver, and deploy, um, agentic process automation solutions.

And so we've created a bunch of learning trails that are specific either to feature or specific to persona. So if you're a citizen developer and you have no background in development, you can take some of our courses and learn the basics, uh, the tenets of problem solving as we call them. Mm-hmm.

Uh, sequence selection and repetition that really forms the basis of every kind of problem solving. And those translate really well into developing and delivering automations. We also have developer learning trails, so that if you are an experienced developer and you want to come to learn to build automations and processes and AI agents, we have training that you can take for that and, uh, learn to build solutions.

One thing that I think sets our training apart from any other training that I've taken, maybe a little bias here, is we do this concept called automation skill stacking, where we'll teach distinct skills in several videos. So, hey, you're gonna learn about conditional statements. Hey, you're gonna learn about logging.

Hey, you're gonna learn about web scraping, right? We'll teach you those individual skills, and then we'll have a project at the end of that to pull all of those skills together in a build. When you do that build, you're doing it in a timed and scored environment, which means we gamify it a little bit.

Sure. You build an automation, you build it in this thing that we've called Agent Quest, where you're building it in that timed and scored environment, and it gives you immediate feedback. So you have the ability to go back and change it.

Maybe you did it with web scraping before, now you want to do it with JavaScript. What does that mean for the accuracy? What does that mean for the speed of processing?

So we've kind of gamified the process of being able to learn and try different things in a safe environment where you're not gonna break anything. Like, the worst thing that happens is it just doesn't work. Start over.

Okay. Yeah. Yeah.

I, I love it. This is a great thing. So, you know, I, one of the companies I helped found over the years was Devox Institute, where we were kind of the leader in DevOps training around the world of, I don't know, I forget how many tens or maybe a hundred thousand DevOps certifications.

And when I came to realize is there are some people who learn for their own benefit, right? I wanna learn, so I know how to do this, it's gonna make me a better developer, it's gonna make me a better, a more valuable asset, more employable, what, what have you. Mm-hmm.

And then especially I see outside the us, um, there are people who are very tied into their certifications, right? I am a, I like to have all the initials after their name. That's right.

You know what I mean? Mm-hmm. Uh, automation Anywhere University.

Can I get the initials or is it just for me to kind of better myself? Absolutely. We have an essential certification, which is kind of that base level.

We've got an advanced certification, and then we have an expert certification. And obviously those ramp up what's required of you. So you start with just being able to complete a, you know, multi-choice test, and then it gets to the point where you abstract to build solutions and submit them, and we're scoring them to make sure that they're following best practices and stuff like that.

Um, and that's something that's really important for us, uh, to make sure that not only are you going through these videos and learning, but you're actually able to apply it and you're being efficient, basically. Right. I would assume continuing education's part of that charter, or, Uh, yeah.

Well, not formally. We, we definitely want to have continuing education for sure. Uh, but it's not tied into universities as of right now.

Got it. Excellent. Um, you mentioned another term, Pathfinder.

Yeah. Not everyone out here again is gonna know Pathfinder. That's Right.

Give us a clue what, what's that about? So, Pathfinder is our umbrella for everything that involves our community, our learning, our training, and our materials and stuff like that. We also have developed something called the Pathfinder Framework.

And this Pathfinder framework is the nine dimensions that we think are extremely important for automation program leaders to consider when setting up and scaling an automation program. So you brought up a great point about training. One of the things that an automation program leader needs to know is have something like a competency matrix developed, which says, I have the following people on my team, and they have the following skills.

Does that help us to meet all of our needs that we have in our opportunity pipeline, so that as we're churning through new automation and AI opportunities, we have the team and the skills that are able to deliver on that? And if we don't, then we need to have some training plans developed and stuff like that. So those are the things that we're guiding automation program leaders on to say, Hey, do you have a strategy and vision?

How are you thinking about governance? How are you thinking about your operating model? Are you federated?

Are you taking contributions from other people of your organization? Or is it just a centralized team? How are you thinking about people and skills?

Your opportunity and pipeline management, your development and deployment, best practices, change management, the way you report on metrics and the way that you evangelize your program? All of those things are extremely important for an automation program leader to be successful. It's one thing just to deliver code to production.

If you're not able to tell people about the business value that is driving for your organization aligned to strategic priorities, you're missing the mark and you're selling yourself short. Agreed. I, I couldn't have said it better myself.

Excellent. Um, a third thing you mentioned was around this whole agentic process automation. Yeah.

Which is, you know, in our last interview with Chief Product Officer Audi, it really is a pivot in the whole, you know, you look at the history of R-P-A-B-P-A, all, all of these things, hindsight's always 2020, let me say. But in hindsight, you look at it and say, this was all great, but it was waiting for something like AG agentic AI to come in and just ignite it, right? Yeah.

I think the most common misconception I hear about is people think that, oh, we did all this work in RPA, now we've gotta scrap all of that. No. Turn to apa.

Right? And I think the, the truth of it is, everything that you've been doing now becomes tools that are available to these agents that can be orchestrated as part of a complex, complex, multi-step process. And so it's really like, Hey, if you have a ton of those RPA automations that you've built, you have gold, right?

Because you can use all of that with those agents, and you can enable them to do things like goal planning and execution, and they can determine in which order to execute those different automations. I think of them as, these are the programs that the agents are gonna run, unfortunately in, in the rest of the world, not in the RPA space. Right.

People make agents and then they've gotta figure out what are the programs? How do you know, what is this agent going to do? And I gotta like kind of program.

Yeah. Here we we're coming at is we already have all the steps. Now I've got an agent that could go do all these steps for me.

Right? That's right. I don't have to worry about stitching together.

That's Right. The, the steps. I think another important thing is you wanna have visibility in one platform where you've got all of these orchestrated processes with agents involved, right?

And if you think about what you have with Salesforce and their agent force, and Workday has their own equivalent, those are great within those specific domains. But what if I had a place to pull them all together along with those RPA tools and all of that capability? And so our platform enables you to work with third party agents as well as agents that are developed directly within our platform.

What a great story. I want to talk a little bit about the Imagine Conference. Yeah.

It's been two days. There's a lot of customers here, a lot of partners, I would imagine A lot of Pathfinder members, a lot of that's right. University students, a lot of the people who you, they're your people.

Yeah. Yeah. Our community is out here strong.

How, how, what, what's been the big story at the conference for you? The energy around a PA and people starting to understand what's possible. And like we talked about being able to use the stuff they've been building, but now do it in a different way, has been, um, definitely energizing and electrifying.

I think another thing that a lot of our customers are recognizing is that those opportunities that they had two or three years ago that they said, it doesn't meet our thresholds for ROI, or it's too complex for us to be able to do with this platform, they can now go look back at those opportunities and start to see like, Hey, we can actually do that now. Mm-hmm. And not only can we do it, we can do it really quickly, and it's gonna be an awesome solution.

So they're starting to recognize and realize that a lot of those things are great ways for them to find their first use cases. I think another key thing is that people are getting hands on with building their own agents. We've got several sessions that we've been doing throughout the conference where people can get hands on building their own AI agents, like I talked about with that agentic quest where it's a time scored environment.

We're teaching people how to do that. The other thing that we're doing is teaching people about how to think about deterministic versus non-deterministic agentic workflows. So I've got a deterministic workflow, it's a bit more robotic, but it's still using a large language model to complete what it needs to do.

I've also got a non-deterministic agent where I'm giving it that set of tools and letting it determine in which order it needs to execute based on the goal that I've given it. I love it. Feedback from people.

You know, there's been a lot of announcements mm-hmm. And, and look at every conference like this, right? A lot of the announcements sometimes are forward looking, and then there's stuff, you know, rubber meets the road today.

What, what has been the, the community, you're, you're the voice of the community, right? What's been the community's kind of take on it? This will be a data-driven answer because we've done an exercise in our community lounge.

There's nothing less. Uh, we did an exercise in the community lounge where we've got this big board setup and we said, what's the feature that you're most excited about that we just announced? And what's the feature that you're most likely to implement first?

Okay. Right. Kind of two different things.

Automator ai, which is our ability to go from natural language to a built automation or process, is the number one thing that people are the most excited about. Right. Think of it as the, the vibe coding for automation.

It's a good way of looking at it. Automation co-pilot is the one that people think they will implement first, which is to say that's the one that's based on Amazon Queue for Business, where you're able to have natural language to interact with your automation repository with those assets, and be able to invoke automations, processes, API tasks, and, and let business users work with them. You know, that that's jives with, you know, text's part of Futuring group, a lot of analysts working with that.

And, and they recently did some, uh, my friend Mitchell actually did a whole report on agent AI for developers. Mm-hmm. And I think that's the sentiment there.

They're not looking for it to be the pilot. Right. Because a lot of people spend a lot of time saying, oh, it generates code.

Is the code good? Is the code secure? Is the code bad?

They're not looking for it to be the pilot. They're looking for it to be the co-pilot. And that, no pun with Microsoft or GitHub stuff or anything like that, but they want AI to be a copilot, not a pilot at this stage of the game.

That may change. Yeah. Right.

As we become, as it gets better, as we become more confident and comfortable with it. But right now, I do think so. It, it, you know, no brainer.

That's the first thing they're gonna do, is take it as a co-pilot For sure. And that connects with all the training that we're doing. We're suggesting that when you get that first pilot use case into production, your first use of an AI agent, you wanna have a human in the loop every single time.

Absolutely. Because you wanna verify every single decision, every single extraction, every single classification that was done by those agents. Yeah.

You can start to peel that back later. But thankfully, our platform has that copilot interface, which enables you to have human in loop a hundred percent of the time. And then you can determine programmatically if you wanna scale that back a little bit and not have every single one for review.

And that way your users are focused on the exceptions or the outliers, rather than the normal use cases that come in every single day. Absolutely. I wanna talk about another thing, if it's okay, Michael.

So, you know, R-P-A-B-P-A-R-P-A more than BPA was kind of, uh, the engine driving the, or, or one of the engines, the major engine driving the whole low code, no code mm-hmm. Citizen developer kinda movement, which is look, has been very successful before there was Gen AI and, you know, it burst on this data. Yeah.

Before this agenda ai, of course, this changes the game, right? We're gonna go from, I don't know, 30, 40 million developers to maybe a half a billion developers mm-hmm. Because everyone could develop.

Yeah. What is, what does that mean for your chart? Or what does that mean?

Like all of a sudden, who is a developer to you? Well, I mean, I hadn't considered those numbers, but the weight of that all of a sudden feels, uh, overwhelming, Right? Yeah.

No, it's a 10 x. Uh, no. I, I, we're really excited about that because A, we've already got a great low code user interface that's web-based, so anyone can just spin it up and use it.

You don't have to worry about setting up your IDE or weird dependencies or stuff like that. It's easy to use. It works in the browser.

It works on Windows. It works on Mac os. Wow.

So we have support for that. I think the other thing is with automated ai, like we just talked about, as long as someone understands their business problem, which is the specialty of business users, they'll be able to explain that and turn that into workable code and solutions that they're able to execute. Citizen development has been kind of a contentious thing in the past, right?

It has. Some people do it really well, some people haven't. Personally, I've led a program where we were really successful with Citizen Development.

Not every single person you train has gone on to build the most advanced automations in the world. But I think with Automator AI and where we're going with a lot of these new agent capabilities, it will enable more citizen developers to contribute in more meaningful ways than we've ever seen previously. Sum Micah, it has been a, a great conversation.

You know, it's an exciting time to be having, especially the Automation Anywhere community you like you did 46,000 people in four weeks, you know, on just Gentech. So it's a great time to be involved in leaning and, and helping this community grow. Keep it up, come back and keep us posted on this too.

It's, it's a, an exciting thing. Um, that's gonna wrap up this one though, Micah. Thank you very much.

We are here at Automation Anywhere's Imagine Conference in Orlando at the Conrad. We're gonna be back with more state tuned. Hey everyone, it's Alan Shimel.

We're back here at the Conrad in Orlando. This is a beautiful, fairly, I think the hotel's less than two years old. It's gorgeous.

And it's the scene for the Imagine Con Conference from Automation Anywhere. And it's been a great conference, two days full of what we now call a agentic process automation. It's really, to me, this is the coming out party where RPA has become a PA and we've been talking about it all day.

We're gonna continue talking about it here with a, he's actually a customer of, of, uh, automation. Anyway. He's also been a key speaker at the event, but more than that, he's somewhat of an expert in this new field of Ag Agent ai.

Yeah. Let me introduce you. I hope I put Tim put too much pressure on him.

Let me introduce you to Rahul Patet. That's right. That's Me.

Rahul is with a company called Alight. Yep. We got that right.

All right. A, you're doing Great. com.

There you go. All right. So Rahul, what do you do at Alight?

So at Alight, um, so first, before I introduce myself, let me introduce Alight to you all. Great. So, alight.

Uh, at Alight, we help more than 35 million people to help access and manage their health, wealth, and wellbeing, uh, benefits. Mm-hmm. And we do this across thousands of clients, right.

Including 70% of the Fortune a hundred companies are our clients. Wow. Okay.

So think it's a huge scale, right. And we have to do a lot Right. To help to ensure that these 35 million people get the right set of benefits for that.

Okay. I That's mission critical stuff right there. Yeah.

For a lot of these people, that's life and death kinda situations, You know, it's, yeah. And health and wealth are the two key integrators. Right.

And then the wellbeing, all these three things. We take care of that. Me, uh, I joined Alight around three years back, and I lead the Automation Center of Excellence for that.

Right. They have an automation center. Yes.

They had it three years ago too. Yeah. Yep.

Wow. So, so I, I, so when I joined Alight, I, at that time, there were three different platforms Alight had, right. Including Automation Anywhere.

And we, at that time, three years back, we were only focusing on automating back office task. Right. Which are all rule-based, deterministic, those kinds of things.

And the Other Well, that's what was possible then. Exactly. Right.

That's, that's what, and, and you're so Right. Right. That was the only thing which was possible.

And there were no, uh, AI and other things were just, just being started. Right. So that's what they were there, and we had 300 bots and more than a a hundred plus team, right.

Maintaining these bots. So I took over a large, complex and expensive portfolio. And, and that's where I decided to see, hey, can I, um, is there, is there a way I, I can optimize that?

And that's where I consolidated my entire automation platforms to a single one saying, and move to Automation Anywhere. So you went from three, let's call them, uh, RR Rrp, three different RPA platforms to a single rrp. So a Single one.

Automation. Yeah. Automation.

Excellent. Why did, let me ask you, why, why'd you, why'd you settle on Automation Anywhere, if you don't mind me asking? Ah, yeah, definitely.

So they, we, we evaluated multiple things, right? So we evaluated technology, we evaluated pricing, and we evaluated customer relationship, right? So what stood out, right?

And actually all three stood out, but the best thing that stood out was the customer relationship, right? They were, there was, and not that time, but today itself, right? To allow, right.

Even when we say, Hey, we need this, they are there right away, right? And then it, we are not a, a backlog for them, right? We are not any other company.

We are, they're, they're giving us the importance they need to from day, from day zero to till date, right? They're being giving us the importance. So it has been a great journey with them.

And that's the reason we chose it. The, the most critical thing was the customer relationship piece, right? Obviously, the cost also mattered, right?

And the technology mattered, but customer relationship was the great things. And I, I'm, I'm, I'm quite happy with it. It, It's so funny.

We're talking about things like robotic process automation and mag AI automation, but yet it's the human to human interaction. Human to human that matters. The trust, you know?

Yeah. The trust factor. And that Lesson there, by the way, yeah.

There's a lesson, right? No matter how fast and how good all the AI gets, we can't forget the humans If we will never be able to. So somebody was asking me, and I, and I, I, in fact, I was talking to either IDC or Garner, I forgot, right?

One of them today morning. And they were saying, okay, we are, we are going edge intake, right? What will happen to those people?

It's not that, right? There is, there is a path, and I have, I've gone through this journey, right? Last year when I came over here, I was talking about a process which we have automated called Claims Process automations, right?

That was a journey At that time. We had an, a huge number of people who were sitting in our back office operations, doing claims processing manually. We had to start that journey.

We started using ai, right? To either approve or deny a claim automatically. And today, I could safely say that, Hey, that bad job.

We, we process more than around 20 to 30,000, uh, claims every day manually. Right? Now, we are using AI to do that, and automation along with Automation Anywhere tools, right?

Uh, to automate that. And we are able to do, we, we are automating all of them, but a straight through processing rate that we are not touching and no human is touching is around 30% of that. 9%, much better than what humans were doing earlier.

So it's, it's a, it's, it's a journey, right? That's what it is. And the new, the new use case, what we are doing right now, it's truly a gentech.

Right? That's, that's what I'll be talking about, right. In some more time.

Absolutely. Um, Raul, when did, in your mind, sort of the, the click happen, the eureka, that, hey, we're moving from RPA to Agentic, pa, right? A PA, right?

Because clearly here at Imagine it's, it's a PA Yeah. It's a PA Last year it was RPA. Yeah.

And this year it is a PA When did that click for you? Uh, see, we are in this, right? We are in a era, right?

Where things are changing so fast. Yeah. Every 60, 90 days, you see some new LLM model coming up.

I was talking to one other partner, and he's saying it's not about large language model, it's large action model now. Right? So new words are being coined, right?

We are talking about ai, right. Somebody else, right? And I think me itself, he was talking about, uh, artificial general intelligence.

A GI, yes. A GI. So think about it, things that, right?

We are right last year, AI things that so drastically if we don't keep pace to that, right? And then it's not about that we'll be left behind, right? The customer end customer doesn't, doesn't matter to them.

Whether I use an ai, either I do it, uh, using HI use, whether I'm using a RPA or an energetic for them, the results are important. Till that time, the results are being shown correctly. Backend doesn't, it doesn't matter how, how we automate that, but it is critical to ensure that we give, see, for us, um, at Alight, our biggest challenge is, uh, or our biggest, our most critical period for Alight is, um, every year, uh, whenever the enrollments happens, right?

All of you, including you, you must be doing your enroll enrollments every year. Sure. Every year.

But before, before that enroll enrollment window opens for you, aam, uh, we at highlight had to go and do lot of complex configurations, right. For thousands of clients. And we have to do it in a very short span of time.

Right. And we need to ensure that whatever configurations we are doing, very specific to your own health, health and wealth, your, your organization's health and wealth beings, right? We have to do it correctly across thousands of them.

So we have an army of people who are, who's doing this testing is very critical. Right? And this testing automation, was there RPA, was there rule-based?

Was there it was not possible. We, we not, it's not that we did not tried automating this testing process, right. But it was never possible using the RPA, the kind of complexities that we have, the kind of variations we have across our customers, across the plants.

Right. It's, it's humongous. Right.

And it was, it was not easy to manage and automated testings. Correct? Right.

But because of this new agent, a KI, right? These new large models, which are coming up, right? So we are using a large language model for text, right.

To, to create test cases. And we are using another large language model, um, uh, and it's called the large action model that we call it, right? And we use the Automation Anywhere engine to create an agent A talks to an Agent B and try and, and, and automating the thing.

That's where it is. That's how, it's not that I choose to use Agent A, it, it naturally came just to answer that. It naturally came that, Hey, That's, I, I, that's the promise of a AgTech is it's not unnatural.

Yeah. It's not, especially when we're talking about RPA. Yeah.

It is the natural progression of, of taking, of Taking to the next level. You mentioned different LLMs. Yeah.

Excuse me. These are LLMs that you created yourselves? No, these are, we, these are all, uh, outta the box delivered by organizations.

We are not, so we are not leveraging any LLM that we are creating on our own. Right. We are.

And, and all of These, but these are not the hype, it's not OpenAI's, LLM or philanthropic or something like that. They are. Oh, they are.

Yeah. Okay. They, they are.

Right. So we are using, some of these organizations are Right just from that. So What do I think there's a word they're using for these now, not the farm or industrial.

Some, there's a word for those commercial, these big LLS. Absolute. They are, man, I, today itself, I heard a new term.

Right. A GI and every time there is A, well, a GI is the holy grail. Yeah.

Right, right. Yeah. You know, when we get to a GI, it's like the singularity, you know, I, I, if I'm alive, we'll see what happens.

But, um, but certainly when it comes to agentic and, and it comes to these, you know, commercial LLMs like that. But, you know, one of the nice things about ag agentic AI is you can rely on that, you know, large LLM large language module, but you could also really tailor more to your specific processes to your, to your world, if you will. Yeah, it is.

Right. And that's where the engine comes into place, right? Yeah.

That's where, see, the automation anywhere a PA engine comes into place. Yes. Right?

That's where the UI agent comes into place that I don't need to worry about what that is. 'cause the engine is working correctly. Right?

Then these Workflow, That, so that's, that's how it's all integrated, right? It's all, it's all, we all have to work together towards that success. Right?

So there is some portion over here, some portion over here, right? There is a technology, there is a LLMs, right? Where are AI and all those things are working together to make this agent a KI process possible.

Yep. Let me ask you a question, if you don't mind, Rahul. So for most of us, right?

We, we, you know, our, uh, benefit elections, you know, choice window opens usually towards the end of the year. Yeah. Right?

Sometime in November, Thanksgiving and closes by Christmas or, or three months maybe. Exactly. Yeah.

As we sit here, it's may right of 2025, you already got through the 2024 rush. Mm-hmm. Right?

How do you think a PA is going to help you for the 20 25, 20 26 Exactly. Collection season. So, See, as I said, right?

And then it's a process. It's the business stakeholders, right? That have to, so these Mike organization, business stakeholders, client people, they talk to, uh, our customers, like those Fortune a hundred companies or those, all those, the thousand plus customers that we have, they have to go and say, okay, hey, customer A, what is the new, right?

What is the new plan? Are you, are you planning to change some benefits that you are providing to your employees? Right?

We have to gather those requirements. The, the first cycle starts right after they have seen Right Once, once the enrollment in window ends, it takes couple of months, right? For an organization, our customer, to figure out, okay, now this is the plan.

I think these are the feedbacks they will get from their people. They'll choose, Hey, this is a newer plan. They will choose.

And that plan, whichever they're choosing, they will come and consult with our business teams, right? That says, okay, this is what we need to do. So we get thousand, different thousand different customers, right?

We get thousand different, more than thousand different requests beyond each customer has multiple requests coming in. And that we now start thinking of configuring into the system. The agent AI process will help us do the configurations, right?

Um, and eventually we are not there yet, right? But yeah, we will automate the, read those requirements, configure it in our system, and then take out what is being shown in our a light work life platform, right? So to ensure that whatever the client has said is being displayed to them correctly.

So that's, that's how the end-to-end process of an agent, a KI will work, and there will be multiple, uh, agents working over there. An agent, which might be configuring it, an agent might, might be executing it, and an agent, which might be validating it. So, so interesting journey over there.

It's going to be an, and this is, you bring up another aspect of this whole agent AI thing, and that is the idea. It's not one agent. There's multiple agents.

Some of them are kind of, you know, ephemeral single use. Some of them will be persistent. Yep.

One of the things I was, I was talking to the chief product officer at, uh, automation Anywhere you could watch the video. Audi. Audi, yes.

About the idea of orchestration Oh, yeah. Of agents. Yeah.

How mission critical is that to you, to using the multiple agents in your scenario? So in My testing process itself, right? We have three different agents right now, and, and we are using an orchestrator over there, and that's what I was talking about, right?

One agent right? Creates the reach, the requirements, which are, it's all in unstructured, right? When the business people or the client team, teams of us talk to our customers, right?

And take the requirements, just take it, right? And they document in a Word document, they put in a table, it's all unstructured. So they put the requirement in an unstructured format, right?

The first agent generates test cases and reads these unstructured data to say, Hey, this is what your test cases are, test scenarios for organization A will be, this will be how it'll be for B, and this is for C, right? That's how it works. So that's one agent, right?

The second agent actually then goes and execute these test cases in our a light work-life application, right? Does the execution autonomously, right? We have this, and that's where this UI agent that we have, uh, the, uh, the feature which Automation Anywhere has given us, it's called the UI agent.

That UI agent navigates through our enrollment flow for each of our clients and their complex scenarios, right? Autonomously, and then gives them the reserve that's, that is, will help us in reducing our testing, takes months, right? If you ask how it'll help us, that testing that month will be reduced to days because of this agent.

Take I and three different agents doing that. The third agent will actually validate it. But yeah, that's how it works.

And You also asked me that, hey, how your 2025 and 2026, I think even though I'll say it right, by, by by next quarter or two, we will automate 80% of this test state. 80%. 80%.

But it's not about technology. My agent a KI technology is great. It'll, it's not about us.

It's about the business, right. And our customers, whether the business will be able to, it's, it's about the mind shift change and the trust that should come in my business teams, business stakeholders to ensure a decision and an action which this agent is taking, is same as they were doing it manually. So I, even though I will complete 80%, but the adoption might be only 20% of I will be happy.

Like even if they adopt 20 to 30% of that, because it's not easy for them, right? And that, that trust factor comes forward. It's a journey.

I think The trust factor is the hardest part of This. Exactly. That trust factor will come slowly and that should happen, right?

It shouldn't, it's, the technology is there, but the trust will begin and it'll help, right? Improvise what the agents are doing slowly. And I think that's, that's what, that's, that's the journey that will happen.

Yeah. That, that to me, that's not a, a question of the technology. That's a, just a question of human behavior.

The Human behavior, right? How was today? How will you just, How fast do we want to trust this?

Exactly. I, that's gonna be, and quite frankly, I think that's gonna be an interesting piece on the whole AI adoption curve. Yeah.

I, we were, we were talking to Micah Smith who runs the community program, developer community program here, and he was saying, you know, literally in one year, ai, you know, specific, specific uses of AI went from like single digits to 65% Yep. In one Year, Because people started to trust it and they start using it. It's the same thing here.

Yeah. I'll, I'll add one thing over there. Right?

Go ahead. Um, what I will say is, see, we are engineers, right? We are automation engineers, but the, our business stakeholders, they understand this process and complexities of our, uh, customers very well, right?

For, if we have thousand customers, we have thousand different business stakeholders for them and a team supporting that. We have a huge army of people supporting our customers. What we will do is we'll create this engine, this agent, a KI platform of testing to them, and we will give it to our customers, right?

So they can write their own testing prompts. It's an English thing tomorrow. Uh, I think down the line, these, the entire surface based automation that we are doing, it'll change, right?

It'll be a plain English that will be doing, and things will be happening. So that thing will go from us, from engineers to citizens back to our business to do those things. It, it'll be a great journey.

What we can think through that. Absolutely. Ul, I want to thank you for coming on.

I know you were a little hesitant coming in here, but I, it sounds like you got comfortable. Okay. Yeah.

Thank you. It's a pleasure meeting you. Yeah, pleasure.

Okay. Thanks. Rahul.

Fitted from, uh, alight here at Imagine Conference, uh, automation Anywhere in Orlando. We'll be back with more, As enterprises roll out production applications using AI model inferencing. They're finding that they are limited by the amount of memory that can be addressed by A GPU and a lot of the other architectural considerations.

This episode of utilizing tech features, Steen Graham, founder of Metro ai, discussing Modern Rag and Agentic AI applications with Ace, Stryker and myself. Welcome to Utilizing Tech, the podcast about emerging technology from Tech Field Day, part of the Futurum Group. This season is presented by soy and focuses on AI at the edge and other advanced enterprise IT topics.

I'm your host, Steven Foskett, organizer of the Tech Field Day event series. And joining me today as my co-host from Soy is Ace Striker, somebody you may recognize from our last season of utilizing Tech. Welcome to the show, ACE.

Thank you, Steven. I'm very excited to be back and, uh, we're here in beautiful Sunnyvale, California today. Thanks for having me.

Yeah. It's, uh, pretty cool that we were able to get together in person to record this episode. We're actually here for our AI infrastructure Field Day event.

And, uh, Heim is gonna be presenting, uh, this afternoon. And so we thought, thought that it would be fun to record an episode of utilizing Tech right here on the show. Uh, talk to us a little bit about, uh, what we're gonna be thinking about today.

Sure thing. Yeah. Well, the, the topic de du jour is AI, as it has been for, uh, our last, uh, several conversations.

Uh, what we're getting into today is really around, uh, the inference side of the AI equation. So we've been spending a lot of, uh, calories there are lately, uh, talking with partners, customers, understanding emerging use cases. Uh, as we've said, uh, there is no AI without data.

There is no data without infrastructure, right? And that's where soy comes in. Um, what we're learning is that the, the sheer magnitude of data, uh, on the inference side of things, uh, is just blowing up.

There's, um, uh, uh, that's not just our point of view. I mean, there's, there's analyst reports from McKinsey and Tech Insights and others that'll kind of reinforce that notion. But what we're seeing is people use these models so much and they, they, there's so much, uh, data involved in going in and outta these models during inference, uh, that it's really putting a strain on infrastructure and it's driving requirements higher and higher at a very fast rate.

And so I'm, I'm looking forward to getting into that topic a little bit today with our guest. Yeah. And, and, and importantly, uh, memory constraints pay a huge, or, uh, play a huge part.

So one of the things you're gonna be talking about today on both the podcast and as well as at the field day event, is how that can be, uh, reduced, uh, through some clever engineering. And speaking of clever engineering, that's why we've got Steen Graham here. Uh, Steen, welcome to the show.

It's nice to have you. Well, Thanks for having me. And Steen Graham, CEO of Metro ai, and we do a lot of work around building AI agents and, and also model and hardware performance evaluation as well.

So talk to us a little bit about, uh, well, I guess let's, let's just go right into it. Um, many companies are trying to deploy AI based applications. Many companies are trying to build applications that incorporate, uh, enterprise data.

Uh, retrieval, augmented generation or RAG has been a huge topic, but what we're finding is that that can place some pretty extreme, uh, stresses on infrastructure and require quite a lot of memory in order to implement in the real world, right? Absolutely. And I think, you know, that the modern kind of transition from your historical CPU first data center to A GPU First Data Center has caught a lot of enterprises in this chasm of that making that transition.

Meanwhile, the pace of change in AI is presenting all these opportunities for them to move to these modern software stacks, but their existing infrastructure doesn't quite work for it yet today. And, you know, I think what we've been working on with the team at soddy is how do we kind of look at those scenarios and pave a pathway instead of going, you know, full GPU centric data center with all your infrastructure, how do we kind of give them a pathway, an affordable TCO optimized pathway into deploying those, those latest AI agents and agentic rags stacks for their their use cases? Hey, Steve, can we, just to start, uh, no doubt some of our audience, uh, is familiar with retrieval, augmented generation, what that is, but can you give us a quick summary, you know, how is that different from just selecting a foundation model off the shelf and, and plugging it in and starting to feed inputs into it?

What, what does RAG buy you and why are folks so interested in it? Yeah, so I think the, the, a simplified view of things is, you know, several years ago with the advent of transformer based large language models, we, you know, you would just use a model, serve it in a chat bot, and then the model would hallucinate. And so obviously for, you know, enterprise quality needs, that doesn't work.

And so the, the next kind of iteration of, of innovation was how do we actually put your data close to that large language model? Um, and notably what sometimes we use a, a vector database or, or a graph database in some scenarios as well. And graph graph could be super useful for, for different different use cases.

Um, but you just basically take all your existing data and usually it would be like, you know, for a particular use case and the, the associated tribal knowledge associated with how that use case is solved. And you're pairing that, that vector database with the language model. So now the language model before it generates an answer is querying, um, you know, high fidelity, accurate domain specific information.

And that's really kind of the simplified view, you know, of, of the rag, you know, environment today. Now, what you'll hear in the year 2025 is everybody will say, this is the year of AI agents. Um, and, you know, I have, I've definitely been hearing that.

Yes. Yeah. Yeah.

And I think AI agents are kind of, you know, an extension of that evolution where we're actually now allowing the AI to query like, and use tools from the company API calls, you know, to the company's internal CRM system or their HR system, or your supply chain system, or your Jira tickets. And so now we can, we can do things like take that domain specific information that RAG already has, add an AGENTIC framework on top of it, and then extensively do that where you can actually create a digital worker that can get the job done while a human's not in the loop. And that's where, you know, people are really trying to look for, where's my 10 x, uh, ROI with ai?

Turns out it doesn't happen when a person's in, in a chat bot, co chatty, and even a rag based environment, traditional rag environments, have had that kind of chat bot type interface with your own data. So you're chatting with your own company's data for higher fidelity outcomes, no hallucinations, but you're still not getting the scalability of a digital worker that the AI agents will provide. Yeah.

And, and that's, I mean, I guess if you wanna talk metaphorically, I mean, RAG is a great idea, but essentially, I mean, it's, it's metaphorically, it's a librarian with a really great card catalog who can look up things and make sure that things are contained within the dataset and that they're the right things and so on. You can validate data. There's a lot of things to love about it, but the problem is it needs to have that really big card catalog or vector database, and it needs to have this huge data set, and that can take up a lot of space.

And that's been, uh, something that's, I think, held this back, even though it sounds great. How do you have that encompass, you know, your, your company's entire corpus of data? How do you Yeah, Well, I think, I mean, most companies, you know, sit on, you know, terabytes or, or petabytes of data.

So like step one is, is just basically organizing that data and the high fidelity data. And there's the, the, like the last 10 years we've went on a data journey. So for the companies that have transitioned to leadership data lakes, you know, they're in a good position to be able to, you know, start vectorizing that data.

And most of the database companies and data lake platforms are now offering the opportunity to kind of vectorize data as well. So there's a lot of opportunity and pre-work that's already been done to put the AI models in a position to succeed. But we're still kind of probably in the position where you do want to nail a particular use case.

So, you know, curating that data set for the pro the domain specific problem you're trying to solve, um, you know, who's the business unit owner that wants to solve that particular problem is still incredibly important. So you don't want to just dump everything, um, you know, into a vector DB and start querying right away. You're probably not gonna to get the highest fidelity results.

And a lot of corporate data is, is greatly outdated, um, as well. So there's, there's obviously some curation you wanna do to set your up self up for success, but there's a lot of pre-work already already being done today to, to make that possible. But that is the most important part of the journey, having your data organized and ready to go.

There's no question there's a lot of data involved in the, in the stuff we're talking about. Um, I'm curious, when you look at the actual, uh, architecture that, that these models are running on, um, and that this rag data is sitting on, can you give us a sense of, um, you know, is this stuff typically done, uh, in, in memory? Is there a lot of, uh, storage involvement in, in real time as, for example, an enterprise is running, uh, you know, an inter uh, an inference workload and consulting, some kind of rag connected external data source?

Mm-hmm. Yeah, I think, I mean, maybe like just looking at kind of the, the models themselves like, and where, where people are battling right now with GPUs, because we're centering the data center and all our applications around GPUs 'cause they're the bottleneck, you know, so any rational kind of throughput analysis always says the most expensive component is where you want to have the bottleneck. And so there's a lot of pressure on the GPUs right now.

The trade-offs that the, you know, the companies making GPUs have is it's really, really costly to put a bunch of memory in the GPUs simultaneously. The larger the model, roughly, the better the performance, the more state-of-the-art model, um, that occurs. So there's this really big challenge, you know, and in the GPU memory around fitting big, big large models in the GPU memory.

Um, and so that's kind of the number one bottleneck that, that we all, we all face, um, in the market today. Now, when you start pairing that with, you know, a rag base architecture, which usually we're running the Vector DB on the CPU and then, you know, we're, we're bringing in all that, that vectorized data, the memory hierarchy kind of levels out a little bit more, you know, more like a traditional, uh, memory hierarchy that you would an anticipate as well. Um, but yeah, the GPU constraints are happening and then you're pushing the workload now into more of a full application systematic software workload that's, that's driving more of the traditional, you know, storage memory as well.

When you shift to a rag based architecture. And AI agents is more, extens is similarly extensible to that where, you know, you're, you're running a lot of application logic for that domain specific use case. API calls that are all happening in more like traditional compute infrastructure that doesn't need to happen on the G-P-U-G-P-U is just focusing on serving that model performantly.

And when you say GPU, I'll just point out too that, uh, when it comes to edge especially, but even, um, increasingly in data center AI and cloud ai, uh, it's other types of accelerators too. I mean, you know, there are, there are definitely acceleration engines out there that, uh, in, in many cases can provide, uh, better service than just a standard GPU, but they have the same constraints that you're talking about. In fact, in many cases, those accelerators have even greater memory constraints.

Yeah, you, you absolutely nailed it. If you look at the companies, um, that are doing a lot of innovation in AI accelerators mm-hmm. I use JS as kind of to, to cover the AI accelerator world almost interchangeably.

Um, but the, those companies in many cases made probably decisions to, to save on bomb costs Yeah. And not have memory, um, you know, in their systems that have, they've driven them to make decisions on how they serve models, and then they have to paralyze model to serving these large models mm-hmm. As well.

And some of them great wrote great systematic software to do just that. But this is like a big challenge in the world today, especially as you, you apply larger models, but also you apply a chain of thought reasoning. We start to like massively increase the amount of inference calls we're doing, um, by giving the model the ability to kind of think through things more.

Um, you know, now you're, now you're really driving a, a significant workload and ultimately a high memory footprint too. Not to mention people are announcing like nearly unlimited context windows, like 1 million token context windows at all. And, and wanting to kind of like make sure we sustain that over time, which, you know, the context window versus rag scenario is, is an, is another, another trade off too ace is because with these massive context windows, you're almost getting rag in that mm-hmm.

MLM application as well. Um, but the kind of the fidelity of an enterprise application, I think still still likes the separation of a rag environment. A lot of those context windows are mm-hmm.

Teed up more for consumer based applications at this point in time. Right. But definitely muddies the waters quite a bit.

Yeah. One of the things we've been hearing about a little more often, and call it the last nine months, is, um, approaches for, for grappling with the increasing amount of data involved in inference, we've seen, uh, storage vendors come to market and talk about approaches, for example, for offloading some of your rag data and, and accessing that directly from storage. Uh, NVIDIA is just a GTC talking about the key value cache and, and approaches for, um, you know, placing that in storage as opposed to in memory, especially as you, uh, involve more complex models or longer interactions between models and that just grows and grows and grows.

Is that a, uh, is that a feasible approach? Is that, is that something that folks should be thinking about as a, as a realistic solution to the problem of memory constraints as more and more data gets pulled into the pipeline? Yeah, absolutely.

I think, um, you know, kind of one of the, one of the gifts that we have that I think is under underutilized today in the AI world, probably 'cause we're all focused on like this GP memory constraint is, um, you know, disc a NN. And what that allows us to do with the, the disc, a NN based optimizations we're allowed to offload workloads onto solid state drives that traditionally would be run in memory. And, um, you know, while the indexing time it takes a bit longer to index it, the net results is your queries per second and performance improved dramatically.

So that's, I think, a little hidden, you know, hack that you can use to lower the, the memory footprint. Um, That's a little probability. I, I wanna, I wanna dig into that one.

'cause that seems a little counterintuitive, right? When you tell someone they can read some data from, from storage as opposed to from memory and you're actually seeing higher queries per second when you take that approach. Is that right?

Yeah. And that's, Yeah, that sounds wild. Yeah.

Yeah. Well, I mean that's, I mean the, I think the work that's been done, um, within the DIS and in working group, and obviously we're spending a little time on the indexing side doing some pre-processing and some optimizations there. But once you've kind of made that, that trade off, when you index that kind of one time-ish trade off when you're indexing, then you know, you've got a little bit of the algorithm based optimization that will give you that, uh, queries per second performance.

And it's not like a two x type differentiator, but it's like same level, uh, you know, level performance, you know, plus or minus. Um, it can even in some data sets be dramatically more. But, um, and then you're looking at same, same level recall accuracy.

So just at a trade off of indexing time. Wow. And so it's indexing time, not even capacity.

'cause I was thinking that it would be a trade off with capacity as well. 'cause the nice thing about storage is that you can have a lot more capacity than you can have with memory. Yeah, that's, that's a fair point.

You're definitely using more capacity with that implementation as well. Uh, but you know, that's capacity that, you know, as long as you're using like the, the state-of-the-art, you know, like PCI based drives, like that's capacity usually have, you know, in the system, uh, in initially as well that you're using for other applications. Yeah.

And that, and that's where I want to go to too. So, um, the side effect of making things, and again, even if it wasn't faster, even if it was just not slower, that's still groundbreaking. Yeah.

Right. And the side effect is that you have much more capacity, and so you can deploy applications with much, much more data to support them than you could in memory, even if it was not the same level of performance. Right.

I mean, even if memory blew it away, you'd still run outta memory pretty quickly. And I mean, so, you know, we've talked about soy, uh, you know, you guys have, you know, very big drives, you know? Mm-hmm.

I mean, I remember the announcement of the 60 terabyte drives and 120 terabyte drives. Uh, I don't think we're talking about having 120 terabytes available to a rag application right now, but are we Well, I think it's definitely, you know, in the scope, it really, I think it really depends on how much high value data in enterprise halves. And if they've got, you know, a hundred terabytes of high value data that's for a domain specific application that improves the quality of the output.

You know, it's definitely, you know, in the scope of deployability today. That's wild. Um, and of course it doesn't just have to be one drive.

I'm a storage nerd. I mean, absolutely. I mean, most, most storage systems use multiple drives, but just the fact that we have that kind of capacity that could be made available to these applications is really, really shattering because there's just no situation in which you could have that kind of ram at an affordable price point if you really wanted to deploy an enterprise application with many terabytes of, of data, you just couldn't affordably.

Yeah. Yeah. So speaking of affordability, one other cool thing we've been, um, having fun with recently is because that kind of core problem that we, that we've always seen about GPU memory footprint, we thought it would be really interesting to see if we can offload the actual large language model onto the SSD.

So this is actually very unique and, um, you know, what, what we've done, And I've heard about people investigating that. That's a really cool idea. Yeah.

And there's some, there's some tools and technologies. Um, in this case we're using a, a feature in deep speed, which, you know, in many cases we use for training applications. But deep, deep speed has some capabilities around model offloading.

And so, uh, what we've done recently is we've taken a 70 billion parameter model. Mm-hmm. Um, which doesn't fit in like a L 40 S based Nvidia, GPU.

Um, and we've actually offloaded the model into solid state drives. And while you don't get the same performance, you know, you couldn't deploy that model at all. You know, and so it gives you kind of model capability based on offloading.

So for people that haven't refreshed all their infrastructure mm-hmm. Or they're waiting to get the latest and greatest GPUs, you can actually use this technique to use a bigger model, um, on a lower cost GPU by SSD offloading. So that's kind of a, a cool innovation.

And I think just like disc a and n has evolved over time and performance has improved over time, I think we'll see a level of innovation and performance improvement and SSD offloading as well, that it's gonna be, it's gonna warrant paying attention to, especially as we're, we're increasing the number of, you know, chain of thought reasoning and applications, and all these inference calls are exploding right now. So at some point you have to look at affordability. Um, and that's, that's a great way to hit a different, totally different level entry point on pricing.

Hmm. The, um, the model off way offload thing is, is really compelling to me. It's a really interesting, uh, idea.

And I wonder, like, my, my, my gut sense is that that may be interesting to folks particularly, uh, who have interest or, or needs to deploy, uh, AI solutions at the edge, because in a lot of cases we have, uh, more severe power constraints, space constraints. You may not be able to put the latest and greatest GPUs in your Edge servers. Right?

Um, do you see that as a, as a potential play for this where, hey, you can now run a 70 billion parameter model on A GPU running at potentially much less power than the GPU would've otherwise needed, and now we can, now we can take that AI to new Edge environments? Yeah, it definitely meets that, that criteria that you look, when you look at the edge, you think about, okay, we're, we're constrained, you know, from power footprint. Um, usually there's a big latency requirement at the edge, um, but the existing infrastructure at the edge that that's lit legacy, typically Edge has a little bit more legacy infrastructure, so it kind of checks all those boxes as far as trade-offs you'd wanna make at the edge.

Now, it might not be a 70 billion parameter. That might be a technique you use on a 7 billion parameter, or if you're deploying on some really legacy infrastructure at the edge, it might be a 700 million parameter model. So I think that it scales down to kind of the, the right footprint, uh, for the edge as well.

I wouldn't, um, ignore kind of the enterprise, uh, cloud applications here as well, because what's happening with, um, with the transition from chatbots to rag to AI agents over time as AI agents are running autonomous of human intervention. Um, now you can always, you know, human in the loop it, but what we want our AI agents to do is they want to, we want them to be our digital workers that are working for us while we're asleep or hanging out with their family and enjoying life. And then we wanna come back in the morning the next day and see the output, all the reports and documents that, you know, the AA agent conducted for us while we were enjoying some great sleep and some great family time.

And that can be done on a batch based processing node. So we don't need to like, you know, get the most high performance, uh, GPU in that scenario. Um, we can kind of use what MacGyver, whatever we have available today, and, and leverage that and then, and deploy it as well for batch based workload.

So I think AI agents offer us a great opportunity to do some trade offs in, in latency. So like real time tokens per second, little less important for AI agents, um, depending on the particular workload. Yeah, we've been hearing that as well.

And, uh, with futurum and some of the research that we're doing, in fact, we're starting to see people talk about using, um, CPUs for, especially for agent ai, for the same reason, because it's sort of an asynchronous workload. Um, also because there's a proliferation of CPU cores. The CPU cores have a lot of specialized functions.

In many cases, they're actually getting specialized AI instructions, and because they have greater addressable, um, memory in many cases than GPUs or accelerators do. So CPUs can look increasingly attractive for this. And especially, and, and, and with many of the things that you're talking about, I could see a lot of that going hand in hand with this CPU trend as well, uh, wanting to use more storage instead of memory to reduce the overall bill of materials to deploy some of these agent applications.

Because essentially, um, you know, you kind of take this to its logical conclusion. We could see SY systems running, um, agentic applications on conventional servers with, you know, a reasonable amount of memory and a reasonable CPU and a reasonable amount of storage, thanks to the fact that we now have capability to use that. Is this a vision that you would share?

I think yeah, absolutely. I think there's a lot of opportunity to take that historical data center architecture and run AI agents on it, whether it's CPU, and we actually have a, a number of AI agents that run a hundred percent on CPU u, new GPU required. Now that being said, I think some of the, you know, older GPUs, fabulous performance still, you know, for those type of workloads as well.

So I wouldn't, I wouldn't start transitioning a hundred percent to CPU in all case scenarios. But for those batch based workloads where you're, where you're fine with a little bit more latency, um, it definitely works. So those existing data centers, I don't think need to be totally retrofitted today in all scenarios, you know, for a GPU centric architecture, um, we can make use of them for, for deploying AI and AI agents.

Uh, one, one more question from mete. I'm, I'm curious, since you're our expert on agents here, and we haven't had one before on the podcast, I want to pick your brain on this. Um, let's say you've got, you've got a model, you've connected it to some rag data, right?

And, and what it can do without a genfy is that the right term without giving IT agency is it can, it can give you insights and advice, right? And then, and then when you give IT agency, you're now connecting it to other tools and systems and allowing it to, to take actions on your behalf. Does the, does the act of giving that model agency have significant, um, uh, repercussions in terms of the amount of data generated or used?

Like you, there's a lot of data clearly involved in training a model. Mm-hmm. There's a lot of data in RAG potentially, and then whatever systems you're connecting the model to have their own dataset, which presumably existed before the connection, but are, is there a, is there a big impact to incremental data simply by virtue of making a model agentic?

Yeah. Is that something you guys have looked at? Yeah, I Mean, it absolutely, I mean, we see this in, you know, our AI agents that have, you know, chain of thought reasoning.

Um, and the more autonomy you give these agents, I mean, the, the human is the bottleneck in the scenario. So if you, the, the better you design the workflow, the more API calls that that agent can do, the more tools it can do, the more, you know, autonomy you can give it and problems it can solve, it's just massively explodes the level of data, uh, being created. I don't wanna like characterize that data as synthetic data, but I would say, you know, kind of non-human generated data footprint mm-hmm.

You know, is, is massive. And I think, you know, obviously I think we're, we're kind of at, at the era where the non-human data generated footprint is, is gonna be, you know, much greater than the human generated data footprint, um, based on giving the AI autonomy as well. And also kind of the advent of modern robotic tools are very data intensive.

So I think, I think that that transition point's gonna be a very interesting transition point. It's great for the storage business, so a, you should be happy. Um, and it's also valuable synthetic data, right?

And, and this is kind of what we're struggling with with AI right now is, uh, you know, there's, you know, the open kind of web data we've certainly ran out of as far as, you know, training models. So then there's a path for synthetic data. There's more human labeling, more reinforcement learning, and now we've got this whole category of, you know, AI agents doing chain of thought reasoning with their data footprint.

And so, you know, all, all of that kind of non-human generated data is gonna be really important for the, the fidelity in the future of AI models. Because if that's high quality data, if if we're generating high quality workflows, then that'll be super helpful. Obviously, if the workflows don't work or the AI model's failing, then maybe that will, you know, devalue that, that type of data relative to human generated data at this time, right?

So it's kind of a really excited to watch Cool. That play out. Cool.

Yeah. Yeah, it really is pretty interesting what's, what's happening here. And I, and I'm actually excited because, you know, generally people, uh, there's sort of a, a thought that, uh, AI applications require absolutely cutting edge, high-end hardware that's really expensive, that consumes tons and tons of power that just, you know, basically there's a lot of negatives around ai and, and many of these negatives are true, but the industry is absolutely working to address those challenges in those criticisms.

And in many cases, we're going to see applications being deployed on much more, um, restricted or, uh, modest hardware with, at, at a much lower price point. You know, and, and I think that all of this means that this, this technology can have a bigger impact than we might have assumed, simply because it doesn't necessarily require the, the, the biggest, baddest, hottest, uh, hardware to run on. You can run it, um, more approachable on, on more modest hardware.

So all of these things, I think, go in that direction, and I think that that's a, that's a positive for all of us. So thank you so much for this conversation. I guess, um, what last thing would you wanna leave our audience with?

What's your summary of, of this message? I think, uh, for me, I think there's, there's a lot of ways to deploy ai. And I think the, the innovations we've seen in the last year are, you know, on affordable deployment of AI are probably 10 x, um, what we saw in the last 10 years.

I mean, it's just an incredible pace of change on driving affordable models. And I think, um, what's most important is, you know, designing the right business workflow, you know, for these autonomous workers or AI agents. And then, you know, figuring out the deployment methodology.

There's so much innovation happening on, on affordable off the shelf hardware, or even affordably deploying a state-of-the-art hardware, um, that I wouldn't let that get in the way of, you know, your company's innovation. Excellent. Well, thank you so much for joining us.

It's been great having you. Um, before we go, where can people continue the conversation with you? ai.

Excellent. And, uh, ACE, it's been nice seeing you again. As I said, you were one of the co-hosts last season.

Um, so check out utilizing Tech season seven. Where else can people catch up with you? Where have you presented recently?

Or, or where are you gonna be? Uh, Boy, oh boy, there's, there's, uh, a lot going on at Solid I these days. Certainly.

com/ai, where, uh, we hope to be featuring some of Metro AI's excellent work in the near future. Um, we'll be at, at conferences all summer long, so, so keep an eye out at, at all the big ones. Uh, for now my head is spinning, uh, with all the implications of what Stein is talking about, and so I need to go have a, have a lay down and kind of chew on some of this stuff.

But really, really appreciate you being here, Ste. Uh, uh, I've learned a lot and, and thank you. Yeah.

And, uh, me as well. And, uh, I will point out that, uh, by the time you watch this episode, the solid IME and, uh, Metro AI presentation will be published on YouTube. Just go to YouTube and, and search for solid IME and Tech Field Day.

And you'll find that, uh, that was part of our AI infrastructure Field Day event, which happened in April. Uh, we're also gonna be doing an AI Field Day event, a cloud Field Day event, and we just announced another AI infrastructure event later in the year as well. So check out the Tech Field Day website for more information about that.

So thank you very much for listening to this episode of Utilizing Tech. Uh, you can find this podcast in your favorite podcast application, just search for utilizing tech, or you can find us on YouTube. If you enjoyed the discussion, please do give us a rating or a review or a comment we'd love to hear from you.

Uh, this podcast was brought to you by Soy as well as Tech Field Day, which is part of the Futurum Group. com, and we are present on the socials. You'll find us on x, Twitter, blue sky, and Mastodon.

Uh, just search for utilizing tech. Thanks for listening, and we will see you next week.