Techstrong TV October 9, 2025

The Nobel Committee goes Quantum. You're watching Textron Gang. Hi everyone.

Happy Thursday, man. This week's flying by Earth Thursday, Mike, you're still, but you're still in Barcelona, right? I am.

Last time I checked and let me look out the window, but yeah, I am. Alright. It's a week in Barcelona.

Good for you. Um, guys, we've got a, a great show for you out there today on this beautiful Thursday. We are gonna talk a little quantum, a little ai, a little, a little Atlassian, Atlassian.

We've got some good people to talk about it with. Let me quickly introduce you to him. I mentioned Mike is joining us in, uh, Barcelona, Mike Ard, happy with the Yankees win the other night.

We've got John Schwartz out in Silicon Valley whose voice is a little sketchy, so we'll try to go easy on them. And of course, joining us, my friend Jeff Rech. Jeff, Jeff Re Jeff is the CEO of the IDSA.

And always a pleasure to have him on here. And we found out today Jeff was a quantum, uh, not quantum, a physics major. Yeah.

School Larger than Quantum. Yes. Much larger.

Billions and billions. But, um, so how Apro Pro to have you on here, so Mike, the, the, the Nobel Committee awarded the other day, uh, their physics awards to three, uh, physicists who did some really, you know, foundational work in the mid eighties that, uh, kind of forms the backbone, the basis for what we call, you know, quantum computing today, With, which I think is maybe slightly overdue after all this time, but I guess better late than never. But we haven't seen quantum computing machines in the mainstream.

At least we've seen them being used in various use cases in pharmaceuticals and some other places. It's interesting and compelling. But Alan, you wrote this article on tech drawing it, and you know, as you were mentioning, it kind of does sort of make quantum feel real.

They got a, they got a Nobel Prize for it. And this is now a, a, a real thing. We are computing with the fundamental elements of nature.

So we've been, we've been computing with the fundamental elements of nature all along, haven't we? Everything comes from nature. But that being said, rumor was they, they were holding off on awarding the Nobel until we did have a, uh, commercial quantum computer, but they were afraid that there would be no one left on the committee.

By the time that happened, they weren't, I'm only kidding. They didn't really say that. But yeah, it is a long time coming.

This was work done as you know, 40 years ago, and hard to believe 40 years ago this was done. And we are still waiting for Q day. That's the bad news, the good news.

Two days closer than ever. So they say, um, you know, we, we have, I I think over the last three to five years we've actually made a lot of, uh, breakthroughs in, in cubits and, you know, all the multi qubit machines, and of course, post quantum crypto cryptography algorithms are, are pretty, uh, widespread these days. But, you know, I think number one, honoring these three men who did this foundational work on, which all everything we're doing now is based, is long overdue and congratulations to them.

Secondly, I do think it's another drumbeat in the steady drum roll of quantum being real and Q date coming third, I think most of our people out here are not really sure other than post quantum cryptography, right? Breaking down encryption, what are we going to use quantum computers for? Jeff, if you don't mind.

Well, sure, I'd love to expound on this because quantum computers, I think in the long run can be used for everything. But, you know, just like every other computer, it's a matter of cost, resources and availability. But I, I think from a of a security and technology perspective, you're gonna see quantum computing further embrace what we do for connectivity be, because keep in mind, quantum particles can pass through objects there, there, there is no physical barrier to them.

Um, now that's an abstract for most people, but we're going to get there with computing, however, and our, our second topic coming up later on with ai, it's directly tied to that because both of them require a whole bunch of cooling and a whole bunch of power of, of electricity. And we haven't solved that problem yet. I would offer, let's target AI and quantum towards solving our energy issues so that we can get more quantum computing.

And Q day can be sooner. But the sort of things you're gonna see coming out of this, I think when you talk about artificial atoms, I think you're gonna see, um, depending on ethical, how it's looked at by different organizations. I think you're going to see gene therapy come out of this.

I think you're gonna see new, new drugs that can target specific, um, issues and ailments and conditions without having other side effects come out of this from a technology and computing perspective. I, I think what you're going to see tied to AI is a faster implementation of whatever it is we need to do with computing. And I know that's a very broad statement, but there really are no bounds.

If you wanna focus on how quickly can we, uh, encrypt something, um, how quickly can we transmit it? How quickly can we calculate what it's gonna be? All of that's gonna come out of it at or around Q day.

I also think, uh, quantum entanglements, which aren't necessarily mentioned in the, um, prize notification, at least I didn't see it when I read it, uh, are going to give us a security, uh, condition that we haven't been able to achieve before. And if you don't know what a quantum entanglement is, and this is some of the work they they did 40 years ago, you, you have two elements. Um, not chemistry elements, but, but two objects.

And they are quant. There is a quantum entanglement between the two. Anything you do to affect the first will have the same effect on the second.

And anytime you make any change to that entanglement, including simply observing it, the entanglement breaks. Now isn't that a wonderful way to have a secure connection that says anytime it's even observed, it breaks. So This, this is where I, I need, I need a little mind expanding, you know, mushrooms or something to help me out here that the, the, the quantum entanglement thing, right?

Because it could be a particle in another galaxy or another, you know, star system, but yet the, the, the entanglement is instantaneous. But of course when it's observed, the entanglement is broken. So how if we can't observe it, how do we know it was there?

And I know they've gotten experiments that show this, but, you know, I stopped smoking dope a long time ago and I just, I can't wrap my head around this. Well, maybe you stop too soon. That's the problem.

There you go. There You go. But, um, if you go back to Heisenberg, who really started the whole concept of quantum mechanics, the Heisenberg Uncertainty PR principle, really that's where it came from.

There is a cat in a box. I know I'm going back to, I think many people have heard this sure metaphor, right? There's a cat in the box, there's a cat dead or alive.

Well, you won't know until you open the box. You observe it, therefore you break the entanglement that was there that said, is it one or the other? Both conditions were true because you didn't know what it was.

And then as soon as you observe it, you know it's broken. You know, which one now exists. I don't know if that helps.

But with the entanglement, the, the good news about quantum entanglements is, as I said, it can, um, they can go through physical barriers, uh, space and distance means nothing. So you can have a quantum entanglement with two particles, uh, different size of the universe. And anything that affects one affects the other.

And any time any change occurs to the entanglement, it's done, which means you've opened the box and looked at the cat. Does that help? Hey, can I mention something that it's, it's, maybe it's a parallel, but what, what I find so interesting about this in this, in terms of what these folks, of these three gentlemen have done in terms of, i, I, I think a, uh, Alan, you mentioned the artificial atom, which bridges quantum theory and, and yes, uh, quantum theory and engineering hardware.

The thing that kind of I see a parallel is a couple weeks ago I went to Stanford. I was at this reception basically about the semiconductor industry and the found the members, the guys who made it all possible. And I think I see this parallel here too.

And I remember at this Stanford ceremony, they were talking about the fair children and what begat the, from the fair, fair children like Intel, a MD alter, et cetera. And I think this kind of marks that same type of moment where you acknowledge the work of the people who build the construction of the house, the foundation of the house, and makes everything possible. And it seems like this was long overdue, but you know, it's kind of gives me the sense of dejavu about what's happened in terms of the long forgotten people decades from now when AI takes off.

So for those who were, um, under 60 Fair children refers to Fairchild semiconductors. Yes, yes, exactly. Yes, yes, yes.

Sorry about that. No, I'm dating myself and Mike, I know in all of us. But I'm thinking about Robert Noy, Gordon Moore.

I mean, these were giants. You all know who they are. Sure.

And I think I wish more people knew who they were, right? And I think these three guys, maybe they will be forgotten, but their place in history will be in the firmament forever. So I have questions.

Am I, do I need all this current AI stuff? If someday I have quantum computing, will quantum supersede ai or will these two things come? No, I, I think, I think AI becomes the catalyst for quantum, right?

And, and, and so, you know, you've got power squared between the two of them that, you know, because for a, some people say for AI to reach its ultimate goal, and if you believe the ultimate goal is super intelligence or a GI or what have you, you're, you're truly going to need quantum kind of, uh, power quantum kind of computing resources for AI to reach that goal. So they actually are very complimentary and feed off each other. You know, I, I did two interviews with the, a, a father and daughter, the daughters of PhD outta Stanford, actually her page PhD's in ai, but she's CEO of this quantum computing company called Q Secure, QU Secure.

And her dad, Dave, who's also very well versed in, in quantum. And the, the, the thing that blew my mind is when you, when you look at cubits, right? And, and which is the fundamental bite of the quantum world, of the quantum computing world, right?

In conventional bytes, computer bytes, it's zero or one, right? So if you have, let's say a 64 bit processor, you have 64 different bits that could be zero or one. So how many permutations of that can you have?

Well, 64 times 64, excuse me, 64 times 1 28. 'cause each one could have two different states, right? 64 to 1 28, a cubit, 64 bit qubit, let's say has more computing permutations.

There's more computing power than what we compute in a year in traditional computing, basically, right? Because it is, because it could be both or neither, right? There's actually three.

So the, the amount of computing power that a quantum computer brings to it, and I, I didn't do this justice. Go back and look at my interview with Dave. I think it was a black hat.

Um, but the amount of computing power that it brings to bear virtually equals the compute, the entire computing power in the world for a year right now. So, you know, what's that gonna do to ai? But if I could offer in, in aeronautics is a term all force and no vector.

That means you have a whole bunch of power, a rocket or a jet takes off, but you're not steering, you don't have navigation, you're just going, to me, that's quantum computing. AI adds that guidance and altitude and direction. Excellent.

I mean, so here's the bottom line though. We've mentioned some of the uses, protein folding pharmaceutical, you know, and, and biomed kind of things. Of course, post quantum cryptography and, and you know, cryptographic kind of stuff.

Um, modeling, uh, uh, weather modeling. And, and, and you know, what, what those kinds of things. I mean, just anything where, you know, you, you, you need that massive amount of, of ability and, and, uh, of, you know, of computing power in, in on the head of a pin.

You know, quantum does that for you. However, let me also say, and I, it's in my article that we haven't solved all the issues quite yet, right? Basically for the qubits for quantum computers to be working now, they've gotta be like flawless or something like that.

And we are developing what they call fault tolerant qubits, which will be easier to manufacture, easier to maintain, right? The work these guys did in 1985 is they had to have things in a superconductor kind of environment at damn near absolute zero kelvin's. And there's a locket junction or something where the magic happens.

Um, I'm, I'm leaning over my head there, but in any event, right, fall tolerant cubits make it a little easier for, for quantum computation to be done without all of those things. There are, Jeff, you mentioned power, the power consumption and cooling and all of that stuff to bring it to absolute zero and all that huge, right? So this isn't the kind of thing you're going to just run because you feel like doing, you know, you want to do some basic trigonometry, right?

You could use your slide ruler. Thi this is, this is an expensive proposition. And, and so the, the rewards have to be equally as valuable.

Does anybody know? I mean, maybe Jeff, like, so how exactly do I write software for this thing? I mean, is there isn't a traditional compiler, right?

So there's gotta be something that I'm using to in access those qubits that Alan's talking about. But what is that? So first of all, it's, it doesn't really exist yet.

So it's hard to say what's actually gonna work. But I believe what you're gonna see is rather than, I think compiler can be a very quaint term when we get to quantum computing, because you're going to, I believe what you're gonna do. And once again, I think AI and quantum are tightly coupled.

You are gonna be able to take a concept and say, here's what I wanna be able to create. I believe AI can create the programming that you need to run on quantum. In fact, I don't think right now a human can comprehend writing a program that can truly take advantage of that advanced speed.

That's why I think AI is gonna have to be the very next step, if not permanent step to doing that. Wow. Alright.

So you're telling me the universe is right here on my thumbnail, perhaps, right? A little, a little animal house there. Animal House, yes.

Yeah. Well, All right. Hey, we're a little overtime though.

We need to take a break. Let's come back. And again, we're gonna talk a little AI now.

Uh, I, big news from, uh, IBM, you're watching Textron Gang. You've earned it. The spotlight, the responsibility, the weight of teams, companies, and entire industries fall on your shoulders.

Lives depend on your decisions, your home life included that work. You are protected physically and digitally. Nothing gets through your team without a fight.

But in a globally connected world, everyone sees you, including those who mean to cause you and your organization harm. And now home your sanctuary attackers see an opportunity. Your digital front door is wide open.

And what compromises your home can breach your board work. Because the devil's greatest trick isn't targeting your workplace firewall. It's convincing you that your personal life isn't at risk.

Black cloak, digital executive protection, defending the new attack surface your personal life. Hey folks, we're back. And yes, as Alan alluded, we're gonna have a little chat about AI this time.

It's IBM making some news with a partnership with Anthropic, and they're gonna integrate the clawed tools into their software development projects, gonna show up in mainframes, all kinds of fun stuff that John covered. But John, um, do you get any sense irony here that, you know, the folks who gave us Watson are now reaching out to get some help from Anthropic? Yeah, I know, I know that, did you know that that's a pretty big, uh, thing there, symbolism, right?

I mean, exactly. IBM claimed that they were the beginning, the forefronts, the builders of AI in a sense that they brought it to commercial use in a sense, they're going to anthropic and Anthropic is gladly gonna help 'em. Uh, so yes, it's an interesting deal in a, in a sense that IBM greatly expands its AI capabilities through a startup.

And it also for Anthropic, I think it's incredibly significant and plays into this narrative that Anthropic has been pushing very hard lately to get into enterprises. Um, there are a lot of companies trying to get into the enterprise through AI capabilities with large language models like cloud becoming more central to business software platforms. In a sense.

Anthropic Hass been making this push into the market of enterprises, um, as kind of the go-to vendor for large organizations. In fact, they've been courting corporate clients since they launched Claude Enterprise in September. They claim they have 300,000 business customers.

And one of the biggest, I think the largest enterprise agreement to date was announced Monday of philanthropic, and it's with Deloitte's. So that would bring Claude AI models to nearly half a million employees at Deloitte. Um, it's also on top of that, anthropic has for ties with Databricks and a partnership they announced, I believe in March, which is targeting businesses looking to develop their own AI agents.

You know, this kind of this divergence between what Anthropic is trying to do. They claim they are now like the enterprise go-to solution versus OpenAI, which is, uh, stressing the consumer side. But, um, it is interesting that IBM chose to work with them given the history of Watson and whatnot.

And, um, I think we're gonna see more of these types of deals between the legacy and between the startups. Uh, it's just really gonna be, uh, almost like a weekly occurrence, I think. Isn't it refreshing to see an AI deal where there's not a couple of hundred billion dollars thrown around?

Remember? So like, Hey, I, I've got my shimmy says this afternoon at two 30, uh, and it, and it, and, uh, it's Shimmy says, uh, when the bubble bursts set to the, to the tune of LED zeppelins when the levy breaks. And, um, and there's also a companion article to this up on Techstrong it that I'll send you too as well.

Great, everybody, you know, to me this sounded a little barish. I don't know if the guys were in purple when they were up on the stage. You know, we're all working together.

It, it is ironic that the folks who brought Watson, you know, the first kind of AI we thought about are now put, put IBM's like that, you know, they'll, they'll put anything in their chin. They got the pipe. If you give them product to put in their pipe, they put it in their pipe.

Um, but here's the real question, guys. I don't want to, you know, stay tuned for my shimmy says this afternoon, but when half of the growth in the US GDP for the year is due to data centers and AI deals half of the go, there's more money being spent on data center and AI deals this year in the US than consumer spending. Consumer spending is usually 70% of our GDP, but there's more of that being spent on data centers and ais as, but here, the, the, the downside is, if you look at the amount of money we've spent on data centers and ais, it roughly equals the entire GDP of Singapore, which is a pretty sizable GDPI think it's 500 billion plus dollars or something like that.

The amount of revenue that AI has generated roughly equals Somalia, okay? Single port is Somalia. com bubble burst, and they haven't lived through the real estate market crashes.

And they maybe have been through the great recession, though we, in tech, we kind of skated that, right? This is a bubble, and everybody's jumping on making Barney announcements and, and pledging hundreds of billions of dollars that they don't have because they're getting hundreds of billions of dollars from another company who doesn't have it. And that company gives it to this company, and this company gives it to that company.

It's Enron all over again. Yeah. You feel you part of Go ahead.

It's gonna be, it's gonna be interesting. We hear these projections of what they're gonna spend on data centers, and it's like more than a trillion dollars of the next four years. And when the reality rubber hits the roads, when we finally get to that point, we're gonna find that it's a fraction of what they said, because it was impossible.

These numbers were just being thrown out there randomly. Well, to Jeff's point, you don't have enough power to run 'em or enough cooling to water to cool 'em, and come on. And that's assuming that we're really going to use it 'cause it's really gonna generate money.

But he, here's an interesting thing, and again, stay tuned to my shimmy says, for this past bubbles were funded by debt, brought down the savings and, you know, the savings and loan crisis, right? com, it was all VC money. This bubble here, it was primarily, at least initially funded by CapEx from the large tech companies who were sitting on billions and billions of dollars.

And now they spent it. But now the second generation of these AI companies, you know, the neo clouds and, and so forth, their, their debt, they're, they're funding with debt. And, and when this thing, you know, when the music stops, those people who don't have chairs, it's gonna be ugly.

Hey, John, are you reminded of the last time IBM partnered this deeply with a small startup company and how that all turned out? So, um, so I'm like, go ahead, Mike. Yes, I know, I know you Kelly.

You know, can I, can I just go back to something that a said, then I'll let Jeff, I'm sorry, Jeff, but it's, uh, in Silicon Valley, the big question is it's bubble or you know, it's bubble or boom, right? Well, there are more people now anticipating the bubble to burst right now. What they're trying to figure out is where are we in terms of that happening?

Are we at 1997 point, or are we at the 2000 point, or are we getting even closer? And, uh, there's several people I've been asking this of, and they're convinced that within 18 months we're gonna find out uncertainty when that happens. Nobody knows, of course, but it's inevitable as Shimmy says.

I, I would, I would agree with that. I I do think that we are, um, it's near, it's near to medium term. It's not a long term issue.

Before this bubble burst for a, a number of reasons. Real estate power, environmental, downstream issues, finding enough talent to be able to manage all this. And then getting to the point of now that we're there, what do we do?

Because it's, we're, we're, it's all forced, no vector. I'm going to use that one again. We're streaming towards this and we don't necessarily have an end goal Right now.

The goal is how much money can investors make? That's, that's what this is all about. And when that, And it's fueling the whole stock market Yep.

And when that breaks, it'll probably break pretty big. Um, It's gonna be catastrophic and, and the thing about it, but here's the real piece that's missing. Where's the killer app?

What is going to generate the revenue to justify this kind of spend, right? I need 10 trillion in revenue to justify a trillion spend. And we're, you know, we we're, we're not sure it really works at that level.

It's great for writing, you know, helping you write, it's great for some marketing stuff. It, it's getting better writing code, but geez, geez, I don't know. Yeah.

There's not a lot of logic. There's not a lot of logic applied to this. It's all pipe dream.

And, and, but you, it's funny, Mike, you're mentioning I IBM working with startups. When you think about Microsoft, I think about even Apple, remember they had that partnership with Apple, with kaleida intelligent, whatever the hell that thing was called. And it, these, these things kind of went awry for Fraud.

You know, the AI train is speeding down the track and companies are afraid of being left behind when it leaves 'em fomo, but they don't know where it's going to end. Is it gonna end at a station or is it gonna end at a, in a mountain at, you know, without a tunnel? No, It's like that meme, I'm sorry, go ahead, Mike.

If you look at the open AI numbers, it would suggest that they are also supplementing the cost of everybody's prompts. So every time you put in a prompt, you know, if they charge you a buck, it's costing them four bucks and they're not Quite figuring it out, they'll make it up in volume. Sure they will.

com days, right? I how many times I heard that in the dot coms, we'll make it up in volume. It costs, you know, I, I charge 90 cents.

It costs me a dollar, but I'll make it up in volume, You know? Well, you hit, you hit the nail on the head when you mentioned Enron. I've been thinking about that more and more recently when I say this, this doesn't, it, it just, it's just like froth.

It's hype. It's over expectation. It's b******t.

And it's, it's gonna come back to haunt a lot of these guys. And it's, it's in it's inevitable. That's all I really could say about it.

I, I wanna put a, a positive spin on this if I can, because it is real and it will benefit us. Yeah, that's true. Yeah.

And it will happen. It's the investment part of it and how much leverage our, our economy has on it, versus is this ever gonna really happen because it will. Yeah.

No, look, you know what, Jeff, it took about 10 years after 2000, but we did use up all that dark fiber in all those data centers we built, right? And we need new ones now. So AI is real, but, you know, this is a classic bubble scenario.

And, and what's even worse is you've got the US government, you know, we're already 35 trillion or whatever it is in debt. You know, our administration's all in on this taking stakes in these companies when no, when those stakes go south, someone's going to, someone's gonna have to answer for santino, Carlos. And, you know, I wanna make sure I understand what you're saying.

You're saying that, you know, because we have AI and we can all now write better emails that we're not gonna move the GDP needle. Right? Right.

Basically. Yeah. You know, pretty much, I mean, here's the flip side.

How bad would things be if we didn't have AI in data centers to spend money and, and juice things up around here? You know, there was, there was an article in the Times the other day that we really have two economies right now. We have an AI economy, which is booming right on the, on this kind of irrational exuberance.

And then we have the rest of the economy, which quite frankly isn't doing so good. Right? And you know, the, this, this is, it's, it's a scary proposition.

It's a scary proposition. And here, here's another thing, and I've said this before, when I hear people start telling me it's a new paradigm, the old rules don't apply anymore. Things are different this time, man.

That's, that's when I keep my hands over my pockets because you're not getting none of my money. Uh, that it, you know, fools, fools in their money. Anyway, go ahead.

Yes, the irrational ex, but this is true. This is irrational. You know what it was?

I think I was younger and dumber back in 2000, right? And I was like, oh, Greenspan, he's putting his foot on this. He's killing it.

I'm, I'm, I'm about to retire, God dammit. But you know, in retrospect, he probably wasn't wrong then, and we're probably not wrong now. Yeah.

By the way, I was in that same boat with you. Yeah. Then, um, but the second thing is, I, I teach part-time.

I've been teaching for a long time. 90% of what I teach hasn't changed in 30 or 40 years. Yeah.

So for every new thing we get, we still have to go back to what are the fundamental principles and do they apply or not? And yes, they do. Yeah.

Agreed. All right. Hey, I think we're outta time on this segment.

Let's come back and we're going to get a report from Mike about what he did on his Barcelona vacation. You're watching Textron Gang, 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 Textron Group. All right, folks, as Alan Sain, I am here in Barcelona at the Atlassian Europe 2025 conference.

They're talking up their robo AI agents that are gonna be embedded into every piece of software they have, and they're only gonna charge five bucks per user. So it's gonna be interesting to see how all that comes together, given our last conversation about the cost and the pricing. But, um, what they are saying is that they're gonna use this MCP protocol that Anthropic developed, along with a graph to integrate all their AI agents.

And every, these AI agents will all know about each other, and they will suggest and help us all do various tasks across their entire portfolio of collaboration. Software includes project management software that people use, JIRA, Trello. We are big users of Trello here and, uh, tech Strong group.

And it's gonna change the way we work. But it was interesting with what they were saying is don't go after big projects. They were saying a lot of the issues we're having with AI is, especially as noted by those folks at MIT, is that a lot of companies went out and tried to hit a home run with ai.

They're saying play small ball. They're saying automate existing workflows, just move 'em forward and, and get some AI muscle memory going and understanding of how to use this stuff and when it works and doesn't work. And that's gonna be the path to success, which frankly struck me as, you know, common sense.

But maybe that is the way to go with all this stuff, is just give it to everybody and they will find ways to use it. And then we'll figure out what the big ROI is later. Alan, what do you say?

You know, they'll make it up in volume. Look, so agentic ai, AI agents, I think are gonna be the make or break of this generation of ai, right? I, I think generative AI has its uses and we've explored them, and we're going to continue to explore them.

But when we look at a, you know, agentic ai, these agents that are gonna, like, let's say cello make using our cellos easier and stuff like that, or workflows that will help us, help us publish articles, you know, on our sites. Mike, it sounds great in practice, but as you know, we haven't been able to get it done in reality. And, and that's where rubber's gonna meet the road.

I mean, and now I, I think this is a classic case of Atlassian, you know, keeping expectations low, saying keep it simple, but it also reminds me when DevOps first started, right? Enterprises didn't adopt DevOps. People adopted DevOps, small teams adopted DevOps, right?

For a long time you had the, uh, Spotify squad model, the Amazon two pizza model, right? And so, you know, it was good for a team of 10 to 12 people. One, you know, there was a huge jump, What a similar message as the CEO of the browser company is here.

And the deal hasn't officially closed between Atlassian and them, but he was saying the same thing about this forthcoming AI browser of theirs. Well, it's forthcoming in the sense that it, well, it works on the Mac today and it'll be on Windows sometime next year. But, um, it was the same thing.

It was, he was saying onesie, twosies, people will adopt this to get rid of their tabs on their browsers and have a more, uh, integrated workflow experience. But he wasn't expecting big companies to kind of sign big deals as much as it was gonna be led from the bottom up. And people are just gonna experiment with stuff and move the ball forward.

I think that too is probably the way to go. And I, I, I think maybe we're just trying to come up with some sort of prescriptive approach to ai, and we should just let a thousand flowers bloom and see what happens Now, who's over 60? I know.

Um, you know, it's interesting. Yeah, this is kind of the same thing that, uh, Atlassian did in, in Anaheim. I remember going there several months ago.

Uh, and it's just a very more practical, reasonable approach. And I think it's, it's the right approach. And I actually think that they, they're onto something.

And, uh, rather than shoot for the sky, let's just see what happens, right? Let's see what happens from the rank and file and see how it grows. And, you know, it's that, that whole mi again, by the way, we, everyone is referencing that MIT report.

It's like just, it's getting torched by everyone, but who, who, you know, who kind of took it, took it literally for what it was. I mean, but if you read between the lines, it was not no immediate effective, uh, profitability instantly. And, and it was, you know, this is gonna take time and agents, maybe this is the year of a i a AI agents.

I kept, I keep hearing that. Now I'm hearing some people say, well, maybe next year, um, we'll see, We shall see. I'm like, I just don't really see yet.

Like, all right, let's say when you do have all these AI agents, and I've got 20 and you've got 10, and the other guy's got 50, and we're all gonna put these AI agents to work, and somehow or other they're all gonna find each other and negotiate with each other, and there'll be some sort of handoff and something magical will happen and it will all get orchestrated. Uh, maybe, but I don't see that happening in the next six months. The crazy thing is like a company, like, I won't mention a company by name, but there are several candidates then announce, uh, AI agent product only to then change and say, well, in a couple more months, here we have a new version.

And this must be driving people on it. Crazy. I mean, how do you keep stay ahead of this, of all these fastballs coming at you, these curve balls?

Many in many cases, right? And, uh, you know, and it's just, it's, it's, it doesn't, it just doesn't make sense. It's just, again, it is baked into the hype, Especially if you've got a CEO who's standing around telling everybody that AI is gonna change the entire economy and the world that we were living and working, and the, the poor IT guy probably looking at 'em going, uh, I don't know what magazines you're reading, but not in my world.

So, uh, you know, I think AI will change our economy as, as Alan referenced earlier, I'm not certain in a positive way in the short term, um, because of the exuberance. Uh, but the, um, but John, to, to your, to your point, so let's go ahead and say we're all kind of admitting our ages. And by the way, I think I win that contest.

But, um, the days of how does it keep up with more fastballs coming at them? They have to, the days of No, no, no, slow down. I can't handle that much.

Doesn't work anymore. Um, you know, the days of maintenance did, who heard of a maintenance window anymore? When's the last time you heard that?

But, but that was the law, you know, 20 years ago. Yeah, That Was, so I think speed and volume are going to increase. They're here to stay.

So we need to adapt. And we, and, and it could be that we use AG Agentic IA to help ourselves to do that. Yeah, no, it's just, it's, I guess, you know, we, I I'll age, you know, I'll date myself again.

When you're talking about Moore's Law, that's like blown outta the water with this, with the, the current, the current environment. Everything is based on speed, adapt, or die, more so than ever before. I think that's what terrifies people.

The, the fact that they're, they're trying to get ahead of something that they can never get ahead of. They're always playing catch up. It's interesting stuff.

Interesting stuff. Anyway, Mike, thanks for the report from, uh, Barcelona. How are the topics?

Is the paella, The food is lovely, as always. The weather's been great. And you know, I, I could live here, You and me both.

Mm-hmm. Okay. Adios from Barcelona, Jeff, John, thank you for joining us.

Thank you for joining us today. As usual, we have full text on TV immediately following our gang episode. And a reminder, again, two 30 live on X and on LinkedIn, I'll be doing my shimmy says we'll dive into this AI bubble business and, um, got a lot more to do.

And of course, we'll be back tomorrow with our week and our week wrap up Friday show for Textron Gang. Until then, though, on behalf of John, Mike, Jeff, and all of the gang folks here at Textron, have a great day, everyone. We're out.

Hey everyone, it's Alan Shimel, and we're back here and we're back live at Swamp Up in the beautiful Napa Valley. You couldn't ask for a better location. The sun is come out, you know, it's good for the grapes.

They say it gets a little cooler, a little warmer, the sun, the rain, you get good grapes, good wine. But we're here with really maybe the highlight of our panel today. Um, three, three of the VIPs of, of the, uh, event, the man to my immediate left.

Actually, I'm gonna let you go last, okay. Let me start to my far left. And I wanna introduce you to Rahul ti.

Rahul is the GVP and GM of the ITSM, something I'm a little familiar with business unit at ServiceNow. Raul, welcome to Techstrong tv. Thank you.

Thanks for having me here. Give our audience needs, no introduction to ServiceNow, but give them a little bit of your background maybe, and a little bit of what you're doing. Its ServiceNow.

Yeah, so I'm relatively new to ServiceNow. I joined little over four months back, uh oh, really? Are new.

And I'm running ITSM, which is the bread and butter, the largest business ServiceNow does. That's where ServiceNow was founded. My background has been building products for a long time for large enterprise companies, but the interesting bit is I switched midway to being a practitioner myself.

So I was running DevOps teams in the cloud space in my last startup before I came to ServiceNow. So I've been on both sides of the equation, building products and consuming products. That's, and that, that's missing in too many of our vendors.

As someone who speaks to vendors all the time, you, it's good to have that practitioner side to see what it is they, they're actually feeling. As that goes by, let's introduce Justin Boitano. Justin is VP of Enterprise AI at Nvidia.

Justin, welcome. Thanks for being on Textron tv. Thank you for having us today.

Give us a little, maybe a little bit of your background. Yeah, well, I've, I've been in Nvidia now actually for 13 years. I guess a little bit of everything over that time, but, uh, you've Seen it come go, Huh?

It's been, it's been, you know, quite a fun ride. Uh, you know, watching us really reinvent how computing is done. Um, you know, from really the ground up.

Uh, and I think it was, uh, when I first joined in 2008, we had just invented Cuda. Nobody knew what it was. We were going around library by library, trying to port applications onto GPUs.

People, you know, thought we were crazy, I guess in the early days. But eventually, you know, every, uh, overnight success is, is 10 years in the making, right? And so we've been working Hard.

That's exactly the biggest secret in tech, the 10 year overnight success. Yeah. You know, we had Avalon earlier, and we were talking about so many people think of Nvidia and they think GP use and the hardware, but the real secret sauce here is Cuda and the software and the ecosystem with partners like ServiceNow and Jfr, and, and that we had Sonar CEO here as well.

Um, and that's really the, the key that's driving all of this is as much as the GPUs do Agreed To my immediate left, immediate left, this man needs no introduction to our audience either. I've had the pleasure of interviewing him for, um, 10 years, 12 years, something long time. He's the CEO co-founder of Jfr Shlomi.

Ben Hayo. Shlomi, welcome. Pleasure being here again.

Again, you, thank you very much for having me. So, look, I was in the keynotes this morning. It was an amazing keynote.

And as one would expect this year in tech, AI was front and center. We've been talking about it all day here. The thing about this is, look, all of us have been around the block.

We've seen technology waves calm and go flow and flow up and down. We've never seen something this disruptive across the entire breadth of, of our industry, right? I, I think Satya Nadella maybe said it best, or the best that I've seen in that we are moving from becoming, you know how Mark Andreessen said every company's a software company.

Yeah. We, were all software companies, but we're moving from software companies to intelligence engines, right? And what, that's a profound change in our business.

Show me if it's okay, I'm gonna ask you to kick it off. What does that intelligence engine bring that to some of what we talked about today here at Swamp Up ai, the whole ecosystem, dev gov, ops, all of it kick us off. So, Ellen, I, I, I think we will need two days to cover this, uh, And then some.

But, uh, but I, I will touch the immediate things that we see in the market. And this is a across of all. It goes beyond sectors.

It goes beyond, um, company size. It goes beyond, uh, geographies. What we see is a revolution, a disruption, uh, that changes everything we knew about the day-to-day practice.

And a company like Jfr, we are not the native AI company. We are the infrastructure company. We are the providers of the peak and shovels.

We are not the gold miners, and therefore we have the privilege to see from below the changes that are happening. So one thing that we see happening across, uh, our portfolio is that consolidation happens not only in terms of technology, but also the inner organization, the CIO and the ciso, the compliance managers, the audit managers, all of them must collaborate in order to overcome the chasm, to bridge it, and to eliminate silos. Otherwise, they will stay behind.

The second thing that we see is that developers, it used to build called, it used to be called, used to deliver software. And then few years ago, they started to be security expert. And now they have to be release expert, and they also have to be AI experts.

0. They are changing the world for everyone. And the last thing that, uh, we see is that if we are not looking at the, uh, opportunity as coexisting, uh, providers, one of us will stay behind.

If we go together, we will change the world together. This is not a race. And, uh, and therefore I'm privileged, I'm honored to walk with companies like ServiceNow and Nvidia, um, to give my customers a better experience, an experience that they expect a one platform experience.

Absolutely. Rahul, I'd like to come to you, right? So you think ITSM, is there any sector of our tech world that is more rule bound that you would think needs a little shaking up maybe, or, or maybe doesn't need a shaking up, but is certainly getting a shaking up with ai?

If you don't mind, share with our audience a little bit of the profound impact that AI's having in the world of ITSM. Yeah. So I think it's getting shaken up.

And honestly, we, being the leaders in that segment, we would like it to shake up. Because if, the way I look at it is, it of yesterday has have changed. Now it is true truly a broker of services, right?

They're managing SaaS assets, they're managing cloud assets. So go on, is the IT of yesterday. The service has changed from IT providing services to I want my self service, right?

So when Jensen was on stage on knowledge, he's like, his main thing thing was, I want my service now, right? That was his ra. So people want their service now and management is no longer like, top down, let me tell you what to do, what not to do.

People are not used to that kind of thing. So we are kind of reinventing it. Service management and AI actually helps you really create that agility on top of the more fixed workflows, because now you can do more with AI to create value out the workflow.

So workflows can still exist. Like the example we gave this morning, compliance and regulation is still required because who wants to have an application that is gonna be breached tomorrow? Right?

That's at the same time. Does it take, should it take a week to get approval? No.

Right. So I think we are reinventing those ITSM processes and kind of integration with jfr, looking at the AI patterns and everything else, because the speed has gone whatever, 10 XA hundred x, right? So things that were taking weeks are taking seconds.

So that's where our head is at from an ITSM Perspective. Absolutely. It's velocity.

It's velocity. Yeah. You know, talking about the profound change, if we, if I asked a hundred people watching this live right now, what is the AI company?

98 of them are gonna tell us Nvidia, but Justin, you know, this, and even Nvidia, I want To know who are the other two, Who the other two, they, they're living somewhere who knows on a, on an island somewhere talking to a volleyball. But, but Justin, even Nvidia knows that as great as Nvidia is and is, they've led the chart help lead the charge here. You can't do it alone.

You need partners like Jfr, like ServiceNow, like sonar, like, you know, so many. You, I mean, one of the, the real strengths here is NVIDIA's ecosystem. Talk to our audience a little bit about that.

Yeah, I mean, I, I think that's, uh, well, a, a great point. Um, you know, Nvidia forever, honestly, has been an ecosystem led company. And I think honestly, it's, it, it comes top down at Nvidia.

Like Jensen realizes the power of an ecosystem for selling our physical hardware through OEMs and OEMs globally, uh, through infrastructure companies, uh, you know, uh, plumbing, the runtimes of these accelerators, uh, up to the AIOps applications and into the GSIs. So we work across the entire ecosystem to try and provide acceleration to, I'll call it, uh, key, you know, workloads that we know are gonna deliver a lot of productivity or performance gains or, um, you know, really kind of transform the, the business, uh, if you would. Right?

Um, and, uh, you know, this, this latest version of it, I mean, for a long time we did it in high performance computing to, to, uh, deal with F-E-F-E-A and, uh, you know, CAE and like the simulation, uh, of the physical world. 0 where it's the, the reality is it's a probably a $3 trillion industry, you know, a a trillion dollars in, uh, pure software that gets bought per year across enterprises and $2 trillion in services. That entire industry is being rethought now through this, uh, this new way of building software where you've got a GenX systems that can break down problems, try and solve the problems on their own, and then reflect on the answers.

And so there's, there's a, a tremendous opportunity, I'd say, for all vendors in this space, really to, to ride that wave with us, uh, in the era of ai. Agreed. If I may add, add to it, uh, Alan, look at, uh, what Nvidia did, um, in, in the history of software, the first thing that they act was a community native company.

They released their software as open source. Yeah. Um, today, um, uh, Justin and his team presented on stage, like everything they build in order to optimize GPU with software is open source available.

That's right. For the community. Open source is not only we build it for you, but we build it with you wi with the community.

So I, I think it really speaks for itself, uh, ab absolutely. But we are looking at the improvements that software brings to the world of ai. Yeah.

com today. My mom used to always tell me, show me your friends, I'll show you who you are. Right?

You probably all heard that. Or a similar thing from your moms, I'm gonna ask, you already said it, Justin, but Rahul, and then I'll come to you. Shlomi.

What's the importance of your partner ecosystem in this brave new world of ai? So, I think any, anyway, at the core of it, if you look at ServiceNow, it is only about workflows data. So that's what we have built our business on, right?

Because enterprise has front office data, back office data, asset data. And if you don't unify the data, then you can be disparate systems on top of it. You tie it with a workflow.

So now the third leg of this tool is AI for us, because AI helps you do your workflows better and faster, and there is no way we can own all the pieces of the workflow anyway, right? There is the software supply chain that multiple players, including jfr, are there from a hardware perspective or from a software infrastructure perspective. Then we have cloud vendors, there are model vendors.

So at the very heritage of the company, we believe that partner ecosystem is critical to it, because that's what our customers want. They cannot rely on a platform that is closed, monolithic does not allow for partnerships. So I think it's the DNA of the company.

That's why we are so excited to partner with. We call it like any model, any industry, any infrastructure is what our ethos is from. Excellent.

Yeah. Shlomi, I, I believe that, uh, um, what our customers are telling us is the, uh, the honest, true and the pain that they experience. And they also know how to put the volume on this pain.

Is it a major pain or something that we can handle? And, uh, every time that, uh, that the technology company is coming with a piece of innovation, the second question should be, what's my ecosystem? And, uh, some people immediately ask the opposite question of asking, am I overlapping?

Am I competing that we are running a platform? We have, I dunno, thousands and thousands of thousands of logos in, in our joint portfolio. For sure, there will be some overlap, but if the one plus one equals more than two and our customers, um, actually ask for it, how can you go wrong?

And when we presented apras the, um, the um, dev gov ops solution we discussed today, we didn't present it as an ecosystem tool yet. It was just an idea in the beginning of the year. And our enterprise customers stopped us right there and said, listen, the people who need to use appt trusts the application owner.

They're not coming to jfr, they're going to service now. And, uh, when we started to, to walk with Rahul team, um, and we spoke with the customers, they actually echoed that. And, uh, and what we've built together and presented on stage here today is just a representation of our, uh, of our customer's voice.

Uh, I'm, I'm very proud to be part of a company that instead of coming as an arrogant vendor, telling them what is right for them is asking the, the customers what will be better? How can I make your life better? I love it, guys.

I gotta bring up a difficult one. It's not on our list, but I've gotta ask you, I've talked to a lot of people about ai, as you would imagine, it's almost impossible to live up to the hype. The hype, the hype cycle is there, right?

And there are a lot of people out here who are starting to, you know, the ankle biters. It's not everything we thought it was gonna be. It's not.

This is a lot harder than we thought. It's gonna take longer than we thought. I think 'cause the expect we set such expectations of it changing the world so quickly, I, and I said this, even if we stop developing AI right now, and we just said, okay, let's digest what we have, it would take seven years to fully integrate it into all of our ecosystems.

But what do you say to the naysayers who are saying, we're not going fast enough. It's not good enough yet we may never get to the holy land to the promised land. Justin, you're, you're Nvidia, I'm laughing and you're gonna go Walk a day in my shoes.

It's moving really quickly. Yeah. Is all I can say.

And I think, you know, in the, the early days of generative ai, uh, you know, people were kind of dabbling. They, they treated it like Java. It was a new technology.

They wanted to upscale themselves, but they didn't know how to apply it to their most pressing business problems. Um, you know, and, and we see now every enterprise really focusing on like, how do I reinvent the core of my business? Honestly, at Nvidia, we've done it ourselves too.

Like Jensen gave us the challenge, you know, double the number of chips that you produce, uh, every other year. So instead of doing a chip every 18 months, do it every year with a design cycle in between. And the only way to innovate at that pace without obviously exploding your workforce, is by using AI and infusing it into the, the business process of the organization.

So we're applying it, you know, to reinvent how we design chips, how we develop software, uh, how we engage customers, you know, through every business function of the company. And we're starting to see that in, in a big way, happen really across every vertical industry, from retail to telecommunications to healthcare. Um, and so I, I think it's moving faster than you might think.

I think, uh, you know, en uh, enterprise in some regard has always been a slow beast. Um, it's always moved. The transitions have happened, you know, more slowly than than we would like.

Um, but in my conversations with CIOs, I think what they realize now is it's time to make that shift. Instead of reinvesting CapEx in standard data center infrastructure, take the leap to accelerated computing and, you know, and focus on building agents that address your core business. And, uh, people are seeing, you know, huge, uh, productivity gains and huge improvements to margins that way.

Excellent. Guys, I got one more question, and it's for Rahul and Shlomi. I want each of you to answer this question looking into this camera.

Rahul, you're gonna go first for all my friends in ITSM, and I'm very good friends with the folks at Idol. My, my friends at People Cert and, uh, Demetrius, and they're out in Greece watching this, but talk to all of the ITSM people out there who were worried, is this gonna take my job? Am I gonna have a career?

I just got into this profession five years ago. What is AI gonna do to my job? So I think my thinking is the train has left the station.

If you get on the train, you will have a job. If you don't get on the train and start kind of on the side, kind of okay with nay saying, I think you may actually lose your job. The reason is when I've seen the successes, people who have embraced, they are having AI do the job they did not want to do in the first place.

Like summarization after closing every incident, really going after knowledge base updates. Who wants to do it? I've seen people who have started going that direction, then they realize, oh, I have not submitted that knowledge.

Why can't I have agent tech do it for you? So slowly they are getting on the train, and the train has gone to the next station. People who are left behind, is it not good enough and all that?

Sorry, that is not gonna work. So let me talk to the software developers out there. First of all, whatever Raul said, amen.

Uh, no, no, no. Nothing to add. Software developers, if you remember the days of CICD that you doubted building software with some tools and automation, if you remember the days that, uh, you said that developers in order to be faster, they need to be a bit dirty and not secure.

Those who didn't, uh, um, jumped on the train left behind, and they're not software developers anymore. You have more responsibility and the next generation trusts you to build it, right? Because we are changing everyone's world.

Absolutely. I'll end it with this. I said it before, I'll say it again.

You're not gonna lose your job to ai. You're gonna lose your job to someone who uses AI better than you. Right?

And amen to that. We've gotta learn. It's a tool.

It doesn't replace the spark, the spark that's in our, all of our brains, right? That create the creator. But it's a golden age for creators.

Absolutely. And we're lucky to be here. Rahul, Justin Shlomi, thank you.

Thank you for watching. We've got, we've got two more. Oh, another day after this.

Another half a day here of, uh, uh, JFR Swamp Up coverage. So check it out. We're on Techstrong tv.

We'll be right back. A MD opens up for open ai, Qualcomm disarms arm data centers in space. Qualcomm picks up a new hobby robot, keep away is Beam buying security.

And we're gonna look at the latest threat to secure enclaves in this week's episode of the Tech Field Day rundown. Hello everyone, and welcome to the Tech Field Day rundown. It is October the eighth, and we are back once again with some more exciting news from the world of technology and some other stories too, because we just couldn't lay off of it.

And we hope that you're enjoying National Fluffer Nutter Day. If you're listening to us in New England, you probably are and are licking your lips. And if you are not, you're wondering what that is.

Be very careful when you Google folks. Uh, but someone who does not have to be very careful when he's doing all of that is my cohost, Mr. Alistair Cook.

Al, it's good to see you again. Nice to be here with you, Tom. And I am always careful what I Google, particularly in a public place.

Uh, one of the things I did Google today was to make sure that I was correct about pierogi National Pierogi Day today. Um, my favorite of the Polish dumplings. And, uh, it's a long way from Poland here, but we have a Polish community who have made me great dumplings here before.

Well, the good news is, is that, uh, you should button down your pierogis and be ready for news from all across the globe, because we have a lot of great stuff coming your way, uh, not just on the globe, but around it. However, we are gonna start with probably the biggest news story that happened this week. And that comes courtesy of a MD and open AI talking about all things ai because Open AI announced that they will be deploying six gigawatts of a MD GPUs to power their future AI infrastructure.

That deployment is going to start with one gigawatt of deployed capacity in the second half of 2026 and play out over the next few years. A MD was also quick to offer a warrant for up to 160 million shares of stock to OpenAI that could lead to them having control of up to 10% of a MD if all of those targets are hit for full vestment, and they make the parlay with that weird, uh, punter kicker combo. But the question that I have for you, Al, is six gigawatts of GPU capacity.

Are we really measuring that in gigawatts now? Is that gonna be enough to move the needle knowing that we have seen massive investments from companies like Nvidia? Yeah.

And even directly with Open ai, NVIDIA's deal with them is larger at 10 gigawatts than this a and d deal. Although if you look at it relative to market share of GPUs, this is a pretty significant deal. Uh, I don't think we would expect, uh, a MD to have 60% of the market share that, uh, that Nvidia has in GPUs.

And so that comparison is an interesting one to see this new growth. I think, you know, there's a much wider context here about all of these announcements, and it does feel like we cover this every single episode of the rundown. Somebody is announcing building vast amounts of data center or vast amounts of power generation or acquiring vast amounts of something.

It just, I, no, I'm not sure that it's news anymore. Only these numbers are huge. Uh, I do wonder whether there's some overlap of the numbers that we're seeing announced that announcing $10 million of this overlaps with, uh, $5 billion of this and, uh, $10 trillion.

Well, we're getting to was trillions of dollars worth of spent? Yeah, trillions of dollars worth of AI spent. There must be a bunch of overlapping in here, and there's possibly some interesting accounting just to make sure that there's new announcements.

Uh, we see this as part of the core weave commitment with, uh, open AI that we saw. I think we covered that last week or the week before on the, the rundown. Uh, that's a $22 billion, uh, worth of commitment the sits at within the, the wider contest of the, the Big Stargate Project to, to bring, uh, hundreds of gigawatts of power for AI over time.

Uh, all of these numbers are outstandingly vast, and I really do wanna see vast returns on those investments because there is a huge economic and, uh, even environmental impact to all of these build outs would better see some benefit to humanity from these, these build outs. Uh, the other perspective is that these big numbers typically are spread over multiple years. These are not things that are happening this year.

As Tom said, the the first turn on for this one is, uh, a mere one gigawatt worth of, uh, GPUs. That'll happen sometime in the second half of next year. Uh, for context, there's a ridiculously large amount of power being used by these GPUs and these racks.

And so we are seeing, um, 200 kilowatt racks as being the sort of power density. And so by the time we got a 200 kilowatt rack, uh, one gigawatt is five racks. Um, that's, is that megawatts?

No, it's, it's 5,000. Sorry, that was, uh, I I missed the, uh, megawatts in between five racks is a megawatts. So to get to a gigawatt, it's 5,000 racks.

Yeah. Okay. That's a pretty big, big environment.

It's, uh, multiple data center rooms. Conclusion on this, there's a lot of money being spent on ai. It continues to be being spent on ai.

People who are betting on this belief we're still very early in generative ai, and that they technology path we're following will continue to require vast numbers of GPUs, vast amounts of compute. Uh, personally I hope we see some innovation that reduces that demand over time. We've cut up the ins and outs of the legal case between ARM and Qualcomm for a while.

Uh, this has been a pretty slow moving law, uh, case of law driven by Qualcomm buying nve. And rather than continuing with their own licensing for the, um, CPUs, they took over the Novia licensing, a move that arm says is actually invalid. And that, uh, arm wants Qualcomm to renegotiate their original deal.

Judge in Delaware has said Qualcomm did not breach the terms and that they, therefore the license through Nivea was valid and owned. Now by Qualcomm, um, of course says we're going to, uh, appeal this and that the, uh, the contract with NIUs ceased to be valid as soon as they're acquired by Qualcomm. Uh, does this sound like a benefit for consumers anywhere?

Or is this just your usual legal wrangling over contracts between two vendors? I think part of it is legal wrangling. I mean, there's only so much that you can wrangle on these things.

I think what is lining up though is that the judge is really taking a, a microscope to this, uh, you know, licensing agreement and saying, why do you think that this violated the terms when this acquisition happened? Because that's not what I'm seeing here. And and it comes back to that idea that we've seen a lot in the industry of, if you buy a piece of equipment or buy a piece of software, is there an implied warranty or a user agreement that goes along with that?

The manufacturers will tell you because they want to get double, they want to get what the original company paid for it, and then they want to get what you paid for. It's like, if you've ever tried to buy like a secondhand device on the internet or something like that, and then wanted to update it to the latest software and been told, oh, no, no, no, you need a new support agreement with us. 'cause whatever support agreement they had is no longer valid.

You'll kind of know what this this means. And of course, if you read the agreements that those devices come with, they're very specific about everything in there, how these are not transferable, how this stuff doesn't work. And some of those clauses have been ruled invalid by courts.

And what we're seeing here is that a judge in Delaware took a real close look at this particular thing and said, well, why wouldn't the rule that Nuvia had with ARM not apply to the company that bought them? And, and Qualcomm tried to argue in court. Well, yeah, that's exactly what we're thinking is that when we bought them, whatever their terms were, became part of our terms.

And Arm is like, but no, that's a completely different thing and they're gonna have to renegotiate with us. And I'm sure there was gonna be an additional dollar amount involved in there, because that's what this really comes down to. This has nothing to do with whether or not Qualcomm is allowed to incorporate N'S data center technology into admittedly a lot of things that are gonna be sent out to the customers out there.

Uh, there, the new Qualcomm arm processors that are going into these Windows IPCs, that's where a lot of that new via technology landed. The problem is, is that Arm wants to get paid by Qualcomm for all of that new stuff that they're developing. And if you read between the lines, which is one of the other things that this story is so fascinating about, you'll recall that one of the companies that came out against Nvidia buying arm was Qualcomm because they didn't want this to kind of line up to be like a massive, um, situation where Nvidia would've basically kind of dictated how Qualcomm's business would've run.

I think Nvidia did okay for itself after missing out on that, but I think Arm is still a little mad at them. And so this is like the whole, well, I can't legally cancel your license, but boy, I can make it miserable on you for weeks and months and years and make you jump through every little hoop. And the judge who already smacked the company down in December of last year is basically saying, guess what guys?

Really stop it. So I I, I feel like there's gonna be some rounds of appeals. Some people are gonna take a look at this, there'll be some more back and forth.

But ultimately, I think Qualcomm's probably gonna come outta this smelling like a rose, which is really honestly the best thing for consumers. Bookstore elects Luther lookalike Jeff Bezos has suggested that he wants to put data centers in the one place uncorrupted by capitalism space. Thank you.

Tim Carty Bezos says that within the next 20 years, he expects to build gigawatt data centers in orbit that will be operated by solar energy somehow though they can't operate them on planet Earth with solar energy. Now, I'm gonna toss this over to the real scientists in the room who have pointed out a couple of important things. One, the massive amount of heat that's generated by your average data center is a lot harder to radiate in space, even if you have all of the copper left in the world.

And all of those components that are gonna be in those orbital data centers are gonna have to be shielded twice as much, because it turns out that those cosmic rays that cause bit flips randomly on earth when you don't have an ozone layer to block those, they're three times as bad or more depending on which math you look at. I'm sure that Bezos remains undeterred by all of US news since this project will finally give him a new base of operations against Larry Ellison's Volcano Island Fortress. But Alistair, I asked my question to you is, are you ready to get a contract to launch yourself into orbit to build a data center?

And what happens if you forget the rack studs back on earth? Well, I think there's a a, there's a really important aspect of this, which is to, to understand that, um, Jeff Bezos is sending things into space all the time. And if it's not rock stars, it might as well be data centers.

Uh, the challenge, of course, is that these data centers should stay in space for more than five minutes, and that's a little bit more than he has yet achieved with, uh, with his space ambitions. Stepping back a little bit, uh, motivation here is that when you look at solar radiation, about 90% of the energy of solar radiation is observed by absorbed by the earth atmosphere. That means that in terms of the solar array required, uh, if you put the solar array in space, it only needs to be a 10th of the size of the solar array on Earth.

That's kind of some of the maybe logic that's not really working here, because as you say, Tom, that solar radiation isn't just the good kind, it's also the not so good kind that damages circuitry. Uh, some of the HPE experiments have seen that in relatively short periods of time and space under two years, um, over half of the SSDs that, uh, HPE shipped up into space had problems. And, uh, that's really gonna be a little bit of a challenge because by the time you add enough redundancy to cope with that, you're adding more mass.

And have you really saved energy if you have to use heavy lift, heavier lift than even the, um, SpaceX, uh, Mars, uh, heavy lift. Uh, if you have to use as yet unseen heavy lift to get these things into space, are you actually using less energy than if you simply left them on the earth? Yeah, Jeff, uh, uncle Jeff, as we like to call Jeff Bezos, uh, is saying that this is at least 10 years away and probably less than 20.

Uh, yeah, predictions about the future are particularly tough. I don't see changes in the basic physics here of heavy lift versus, uh, the, the energy that you're saving by heavy lifting being beneficial. Uh, maybe there's something around being in low earth orbit and and leveraging technology like starlink to get this as being the biggest edge compute deployment.

But if it happens, I will be absolutely stunned. Although, unlike Peter Beak, I will not commit to eating my hat if this heavy lift happens. Peter Beak, of course, is Rocket Labs, my fellow countryman, uh, spaceman.

It's time to break out your breadboards. Um, actually I've got a big pile of breadboards just over there as well as some Arduinos, because Arduino is gonna be part of Qualcomm, or at least is gonna be owned by Qualcomm. They promise that, uh, Arduino will still be operated independently.

The acquisition, of course, was announced in the home of Arduino to run in Italy, along with a new board. The Uno Q The Uno Q is maybe a competitor to a raspberry pie that runs a real Deion, not a fork. Qualcomm says The move is exciting way to get into the developer community to walk, augment its business to business focused, uh, chip set development.

Uh, the Uno uses the new dragon wing, QROB 2210 ship, uh, along with a microcontroller on the back. So that combination of high compute power, probably with some ai, uh, chops in that dragon wing, along with, uh, a lot of, with a realtime interactions through that. Um, microcontroller looks to be an ARM-based microcontroller.

Uh, interesting to see that combination together on a board. And interesting to see that as both a standalone computer and, and a development machine. Tommy, you gonna trade on your laptop for an Uno queue?

Uh, probably not yet. Uh, mostly 'cause it isn't come with a monitor, but I think it's interesting that Qualcomm basically kind of jumped right in here with both feet and admitted. Why they needed to pick this up is because nobody buys Qualcomm equipment to develop on, unless you're doing stuff directly for Qualcomm.

And we've seen this for years, right? Like, I still have a pile of raspberry pies sitting in the corner, uh, doing very random things like I think one of them is emic shouldn't machine, and, and there's some other stuff, but it was really cheap to get into developing for arm microprocessors by doing that, right? What was it?

You know, um, 30, $40 us. And that meant that a lot of people were writing software to run on raspberry pies. But as you mentioned, raspberry Os is not Debian.

It's a custom version of Debian because it's optimized to run on pies. And so Qualcomm was like, well, on the one hand, we've got people who really should be developing for our platform. And on the other hand, we've got this really successful group of small device manufacturers who are trying to effectively create this, uh, marketing arm, if you will.

So how can we do that? And when you're Qualcomm, the answer is, I don't know what's for sale. And so they did, they picked up Arduino and they basically said, you know, we're gonna keep it independent.

We're not gonna stick our fingers in it too deep. Um, I, and we're gonna come out with new hardware, right? Um, the, the smart thing to do probably would've been to partner with Arduino first to see if people were actually gonna buy the Uno QI don't know if they will or not, but I mean, ultimately this can't hurt Qualcomm, because if you can start putting Qualcomm, uh, dragon wings, snapdragon, snap wing, dark wing duck stuff into the community, then what you're gonna get is people who stop writing code that runs specifically on pies and starts running it on the, you know, queue.

And by the way, shout out to them for not making this thing a hundred bucks. Um, the, the price points on these things are actually reasonable. Uh, you can get like the, I think it was the, the 16 gig version that's available today, the 32 gig version versus eight and 16.

I forget which one it's, but there's, there's a small winning by today and there's a little one that you can buy soon. Um, that is a great way to get people to start writing. And more importantly, the microcontroller aspect of it is directly aimed at Raspberry Pie people, because having that, that, that the way that the microcontroller works on some of those boxes is basically saying you can swap these things in and start developing for us.

And then later when you want to graduate using big boy stuff, well, it's still just Qualcomm chip sets into the hood. So you can already run that. And are you comfortable developing on, on, you know, Debian then?

Great. We, we run Debian, not some customized version. And, you know, I I think it's a win ultimately for Qualcomm provided, you know, there's a little asterisk there provided they really do keep their hands off of Arduino, because one of the things that we've learned over the years about the hobbyist and development community is if a big corporate entity jumps in and tries to basically impose their will, we are going to see people defecting away from Arduino as fast as possible.

And they've already had their own challenges over the last few years. It was things as simple as who owns the trademark for the name. So the, the, the community's already kind of a little bit wary of what could happen here, but I'm hoping for the best at this point.

All right, keep your distance chewy. Just don't look like you're trying to keep your distance. And that is according to Rodney Brooks.

And you may know the name because he's the founder of iRobot and an MIT professor Emeritus. He says that people should stay three meters away from any of those modern bipedal robots that you see out walking around. You're probably asking yourself, well, Tom, why should regular people stay 354 barley corns away from these walking terminators?

Well, that's because of physics. And our friend Mr. Brooks says that modern robots are expending so much energy trying to stay upright and stable, that any catastrophic failure of those balance systems are going to force the energy to be expelled somewhere.

Remember folks, physics is a harsh mistress, so that means that energy is probably gonna go somewhere into the nearest pedestrian. Brooks goes on to say that current robots cannot develop enough dexterity through vision only AI systems. They're gonna need access to other data such as touch.

Now, I will say though, that in the article, there was no mention of the minimum safe distance that you need to stay from away from your robot vacuum. But al do you think that it's probably a good idea to stay away from anything looking for Sarah Connor on the street? Well, It depends what your name is.

Uh, if your name is Sarah Connor, then definitely stay away from, uh, whichever kind of, uh, walking full-size humanoid things are out there, particularly battery powered mechanical things that maybe can't feel, both can't feel physically, but also can't feel emotional. There's some interesting challenges in here. Some, some, some thoughts around this.

One of the elements is that we kind of like our robots to be actually androids. Things that look like humans have human form. Uh, yet there are more efficient ways to move around.

I was reading about, uh, putting robots into hospitals. Well, hospitals are designed to have things that roll. So there's no need for your, uh, service robot in a hospital to have legs, and therefore you can get a lot more stability because there never needs to be the, the ability to manage a single point of contact with the crowd.

And I think this is kind of where Rodney Brooks argument comes in from and why he's not afraid of his robot vacuum, because it's not standing six foot tall, one legs with maybe 200 pounds, maybe 400 pounds of total mass. I haven't looked at what these things weigh now, but I imagine they weigh a little more than I do. Uh, his concern is that when things go wrong, they go wrong in a catastrophic way.

And that kind of was born out by his early experience of being a little too close to a humanoid robot when it did experience a failure and went down. Uh, he's still alive, so it wasn't too catastrophic. But I'm not sure I'd wanna take those risks too.

Uh, for context, this is, uh, three meters that's 10 of your imperial feet, or just over three yards. Or as you say, what was it, 480 barleycorns. Um, stay far enough away that when it falls down, the bits of fragment that go flying outwards don't hit you.

Or if it's arm goes up in the air as it's going down, you get, don't get slapped by. It was the suggestion. And really you come back to why do these, uh, things need to have two legs?

Why can they not move in a a different form? And why do the two legs need to be placed in the same way that a humans are? Uh, these robots would probably be more stable if they had three legs, but just how far across the uncanny valley would we be if these things with human voices and arms walked around on three legs, or even possibly four, so that there was always three in contact with the ground for stability?

All of these things would make these, uh, robots easier to design, easier to operate and consume less power. But they don't look quite as human. I'm not sure if I need them to look human.

Maybe I'm happier if they're safer to be around. And maybe the, uh, kind of blocky things that we've seen in the past might be a better move than some of the things we're seeing from companies like, um, uh, like Elon Musk, uh, plans to have these blank faced humanoid robots serving us our drinks. Mm.

I'll give them some space. Veeam Veeam's ready to make a big splash in the data security posture Management. 8 billion for security with an eye well with two eyes.

In fact, uh, this would help them compete with Rubrik and Cohesity more directly. Hmm, backup companies getting into security data protection. Uh, DSPM is a big driver for security teams right now, particularly looking to manage the security of data that's being used by AI applications.

Uh, Veeam, of course, uh, having moved into private equity has been making a bunch of acquisitions, uh, particularly, uh, picking up the ransomware negotiations group. Core ware, sorry, Cove. We, uh, not core.

We, uh, ransomware negotiation group, core ware, I've got that wrong yet again, I'm tripping over all sorts of things. This could be another in a recent spate of acquisitions from Veeam with most recently picked up ransomware negotiation group, Cove Ware, uh, Tom, is this shifting Veeam from, well, my kind of data center infrastructure world into your security world, is this significant for Veeam? It is.

And we heard from Veeam at Security Field, a 13 earlier this year. Uh, Rick Vanover, uh, and Emily did a great job of kind of presenting about ware. But if you look at where companies are now, it's not about those things.

It's about creating a holistic view of security and, and what you said, uh, data protection. And one of the pieces that Cohesity has had for a while at Rubrik has working on at Veeam is in need of development, is data security, posture management. And essentially what that is, is going through your organization and finding out where all your data is and figuring out how to classify it so you can assign risk levels to it.

But there's a lot of other benefits. And one of those is that if you properly classify data, and I'm not talking about like government classification, where it's like, you know, secret, top secret, and I could tell you, but I'd have to kill you. Um, this is about, you know, how sensitive is this?

What is a risk factor associated with it that's super valuable for ai, right? Because when you look at the way that we're using AI to do these kinds of things right now, there might be some stuff that you really don't want to have included in your algorithms. And being able to classify that and provide a posture to it and say, okay, anything that's classified as, you know, label X never included in the algorithm, is gonna be super valuable.

And companies are gonna be looking for, I'm sorry, customers are gonna be looking for companies that touch that data. Are you gonna go download a tool that does that, or are you gonna rely on the company that already does look at all of your data, which would be your disaster recovery data protection company? I know which one I would rather give permissions to do that.

And so this is Veeam basically looking through the whole market and saying, this is an area that has a lot of synergy with what we're doing, which means we need to be in it. And in order for us to do that, we need to acquire a company that's already kind of a leader in that space. And this goes back to the comment that you made that, you know, about five years ago, Veeam got bought out by private equity.

And I, I don't think I'm preaching to the choir here when I say that the way that private equity solves problems is with a checkbook. They're not gonna look to develop this technology in-house, or if they did, eventually they'll hit a point where the return isn't worth what they're spending on it. So the next thing they're gonna do is they're gonna go out and buy somebody.

Now, again, this is not a done deal, isn't even an announced deal. This is a rumor that this could potentially happen. But one of the things that I love about these kinds of rumors is that sometimes they are trial balloons, because if the entire industry starts nodding their head and looking around and going, yeah, yeah, that would be a good pickup for a company like Beam that's basically being, getting market validation and in some ways causing valuations to rise.

I'd say stock price to go up. But we know that's not the case. But you know, if, if, if, if the industry as a whole says this is a good move, I think that's something that they're going to go with.

It's now time for a closer look. Security researchers have figured out how to break enclaves. Hmm, sort of in theory in papers published by two independent groups, um, using the latest exploits.

Now, security enclaves is a way of keeping secret data secret inside your application secret, even from the underlying platform and encrypting data at rest. Uh, both of these attacks called battering RAM and wiretap work by installing an interposer between the RAM and the actual system and capturing the data and replaying it later on. Uh, you need a specific hardware device, you need a specific, uh, application, uh, usually a virtual machine to do that replay at this, the particular location.

Tom, does this mean that trusted enclaves where we put our most secure, secure data aren't secure anymore? It depends on how you look at it. So here's the thing, like this was a, a big story, right?

Oh man. Secure enclaves can totally be hacked. And then you dig into it and you realize you had a long way to go from the TPM chip in my iPhone being safe enough for me to unlock it with my face to like levels of like, you know, uh, Shaw one or, or MD five hashes being broken.

So first of all, you have to understand that both of the secure enclave, uh, compute mechanisms, intel calls, there's SGX and a MD has one. They call S-E-V-S-N-M-P-S-N-P, not SNMP, that's a different thing. They both use something called deterministic encryption to store data in ram.

What does that mean? That means that if I encrypt data on one side and transport it to the other side, the cipher text is always the same, right? So like, you know, think, think of something as simple as a ROT 13 cipher.

Whenever I encrypt it over here, whatever shows up over here with the same key is gonna look the same. But you're gonna say to yourself, that's exactly how data is supposed to work, right? When you encrypt it, if you jumble it in encryption in the transport, it's not gonna come out looking the same.

The reason why they did this is for performance reasons. And the researcher who did some of the initial work, I believe it was on, uh, battering Ram said this. He said that originally Intel allowed SGX to run on consumer processors and enterprise, uh, processors.

And then about four or five years ago, they said, we're not gonna let it run on consumer processors anymore because they were, uh, RAM limited, uh, to around 250, 60 gigs of ram. Whereas on an enterprise, you know, XON thing, we can throw a RAM at it until it falls to the, the core of the Earth. But the catch is, is that in order to scale performance to that level, they had to use deterministic encryption.

You can use non-deterministic encryption, but it does include incur performance penalty. And now IT act being able to use non-deterministic encryption would invalidate this attack completely because you can't replay non-deterministic encryption because the, the cipher text is different every time you use it. That's the mechanical thing.

Let's talk about the physicals part. So in order for this to work, you have to have physical access to the hardware device and it's, you have to install something in the system called an interposer that literally sits between the RAM and the CPU and it intercepts all of the, the, uh, processing that is sent to ram. Then you have to watch a specific memory address for ciphertext, and then you can decrypt and replay problem.

You have to have physical access to the device, right? And I will recall one of my favorite Microsoft quotes of all time where if you pop a Windows xp or if you pop a Windows two Windows 2000 server CD into a Windows 2003 server and boot it, you get unfettered administrator access to the recovery console. And Microsoft's response when this bug was filed was, if you don't have physical access to the box, you don't own the box.

They do. So I could see possibly that a supply chain attack could, could cause an INTERPOSER to get put in there, but that's an awful risk, folks. Like you've be targeted in order for this to happen.

So you've got, you need to have an interposer. It relies on the fact that SGX is doing this for performance issues, and really all they need to do is turn on non-deterministic encryption and this whole thing goes away. And battering RAM works on a MD and Intel wiretap only works on Intel.

It is very specific to using Intel. SGX. I don't wanna say there's nothing going on here because obviously this is, you know, this is like spectrum meltdown, right?

At first we're like, oh yeah, pipeline branch execution prediction, that's a problem. Um, and it ended up being something that caused a lot of problems. But the last I checked, I didn't recall anybody using Meltdown to steal my credit card data.

I mean, Al do you think that having physical access to the box is a bigger problem than putting an interposer in there to possibly inter encrypt intercept ciphertext and an encrypted enclave? Well, If, if these encrypted enclaves were on people's laptops, I would absolutely be concerned when laptops and desktops, I'd be concerned about this because physical access is much easier to achieve with a relatively portable device, physical access into a cloud provider's data center or an enterprise data center that should have good physical security because without physical security, you have no security. Um, that's less of a concern for me, uh, in terms of the viability of a supply chain attack to put an interposer into every single motherboard because this would happen either at the motherboard level or potentially inside the actual dim that is, uh, being, being installed in these servers.

Yeah, that's not a credible thing, but if there's enough to be gained, a nation state actor might achieve this, would never hear if they did, because anybody who knew about it and leaked the information would be, um, unli. Uh, and so, so in terms of the, the likelihood of this occurring, my credit card details being stolen this way, I don't see that as particularly likely. Um, the kind of crown jewels that would be attacked this way would be designed for nuclear deterrent submarines or those kinds of, uh, central banking records that are stored in on these secure enclaves.

But these secure enclaves are just another layer of the security onion. They do not stand alone. They stand in the start with physical security.

Start with single tenancy, because part of the replay is that you need to be replaying at the same memory address in order to be able to decrypt that data. Uh, so there's, there's a collection of circumstances that good security practices will allow you to mitigate any risk that is here. Um, and that's always the reality of, of a security situation is that, uh, you look at what the risks are, you mitigate the credible risks and the cost effective risks to mitigate that are less credible.

Uh, this is probably a, a risk that is fairly well protected by the majority of organizations that are using trusted enclaves. This isn't to say people that are bad at it are not using trusted enclaves, it's just that, well, if you're bad at it in general, you can be bad at it with a trusted enclave and pay a lot more for it. Uh, you know, that's, that's the reality of the world we we live in.

There's that any large valuable target is going to be, uh, attacked in, in multiple ways. And this is just another potential way. We are a little way away from it being a real danger to causing real problems in, uh, in organizations right now.

I couldn't have said it better myself. Uh, you know, your bank account information is not very important, but a hundred people's bank account information is pretty critical. Or the encrypted wire transfer data that you could then substitute your own account information in there, that's valuable.

So I'm, I'm gonna wait with cautious beta breath to see exactly how this plays out, but for now, it's a neat proof of concept and apparently you can build a wiretap, uh, system for less than 50 bucks. So who knows? But I can tell you something that's way less than 50 bucks.

It's free. And that's the upcoming field day events that we have going on. I have one going on tomorrow, starting at 7:00 AM Pacific.

We're gonna be hearing from our friends over at Microsoft. We're gonna be talking about Microsoft security and specifically Microsoft Sentinel. You probably saw last week they had some big news come out around Microsoft Sentinel, how it's evolving, some of the cool stuff they're adding into it, like, uh, you know, MCP server and a data lake.

Well, if those are things that are interesting to you, make sure you tune in. Uh, like I said, Thursday, October the ninth, that's tomorrow. And, uh, we're gonna have some great conversations with some of their executives like Scott Woodgate.

We're also gonna be getting some demos and we're even gonna hold a round table. So that should be real exciting. And then next week we're gonna be out in Vegas, or at least FoST is because he's gonna be at NetApp Insight, which I believe is one of three or four shows that are going on in Vegas at the same time.

It's like they're trying to pack him in before Thanksgiving, or some big thing that's happening in Vegas after Thanksgiving. Whatever it is. Make sure you tune in on October the 15th, because Steven is gonna have some great presentations.

It's gonna be a really exciting time. I'm sure he will have a joke or two about Vegas. You never know.

Uh, but after that, Al, you're back with more great stuff. Not in Vegas. Not in Vegas.

I will be in San Francisco in, uh, glorious San Francisco for Cloud Field Day. October 22nd and 23rd. We have a great lineup of presenting companies.

We're returning to Oxide computing, and we'll hear from Pure Storage and, uh, a couple of other great companies and a, a great collection of delegates as well joining me out. Um, so tune in for that October 22nd and 23rd LinkedIn or on, on the Tech Field Day website. Also text on TV the following week, we hand back over to Mr.

Foskett. Steven will be also back out in California for AI interest. Uh, AI Field Day, sorry, AI infrastructure is my event.

And that's not until next year. AI Field Day will be in, uh, October 22nd. Uh, AI Field Day will be October 29th and 30th.

And Steven has another great collection of companies looking at the things that you can actually do with ai, maybe deliver some value to your organization. Something else that delivers value to your organization is the Tech Field Day rundown and the news that we bring to you every week. You can catch new episodes of The Rundown every Wednesday on YouTube or in your favorite podcast application.

Or if you find us, please give us a like, and maybe a nice review. You can also find us on Techstrong tv as well as you can catch many of us, Tom and myself and and Steven across Techstrong and Futurum Group programs. We will be back next Wednesday with all of the news that's fit to print, and some of it may involve ai.

Uh, until then for myself, for Tom Hollingsworth, and from all of us here at the Tech Field Day Family is wishing you and your family an awesome day. Welcome to another episode of the AI Security Edge, where we explore the intersection of cybersecurity and artificial intelligence with the leaders who are shaping the future of digital defense. I'm your host, Caroline Wong, tech Strong TV podcast features your favorite video series, industry thought leader commentary and analyst research on DevOps, security cloud native and digital transformation.

In a podcast format, AI is revolutionizing cybersecurity, both as a weapon for attackers and a shield for defenders. The AI security edge dives deep into the evolving cyber battlefield where AI-driven threats, challenge traditional defenses and cutting edge AI solutions offer new ways to fight back. Our podcast explores real world case studies, expert insights and practical strategies for building cyber resilience in an AI powered world.

Whether you're a security leader, practitioner, or AI enthusiast, we hope you'll gain valuable knowledge on the risks, innovations, and ethical considerations shaping the future of digital defense. Today's guest, let me see if I can say this right. So there's an American version, which is Francesco Sipe, but then I'm gonna sip poona, sip.

I tried. And then, and then, and then the proper version. We're gonna try Francesco Chip.

Yes. Boom. Way better.

Okay. I'm like extremely proud of myself for that, but you know, just, I think that might have been like a one time thing. So we're gonna call you Frank.

Frank, thank you so much for joining us. That's Frank. Frank is a cybersecurity leader, entrepreneur and thought provoker.

He is at the forefront of application and cloud security. The most important thing that you need to know about Frank is that he was a practitioner, and now he's a CEO. He is the founder and CEO of AppSec Phoenix, also known as Security Phoenix, a company that is pioneering contextual, risk-based vulnerability management from code to cloud.

Frank has done all sorts of cool stuff at HSVC, at AWS, at the uk and Ireland chapter for Cloud Security Alliance is a professor at Ions. He is a multi award-winning podcast host. It's actually weird for Frank to not be the host right now.

He's a regular keynote speaker, he's an author. He writes books, white papers, articles, and, uh, he's also a self-taught artist and a former professional skydiver. So if this is the first time you're meeting Frank, I'm so excited for you those though, because you know what, Chad, GPT, uh, there, which is actually like, that's an AI use case, right?

Um, if this is the first time you're meeting Frank, or are you in for some good stuff because there's so much good stuff, Frank, welcome Caroline, as always, you shine. Thank you for having me. So Frank, we should find all this stuff, everyone.

Uh, literally it's chat, GPT. So everyone on this podcast, I, I like to ask the same questions, but, but you're not like a, you're not like a typical podcast guest. Not really.

And so I'm gonna ask you a different question, which is tell me what you actually really think about all this AI stuff. Tell me the real brutal raw truth. It's a bubble.

Uh oh. Uh, but it is a cool Bubble. Okay?

Okay. Tell us more about this bubble. com bubble.

com or not as experience. Instead, right now, we have organizations still trying to figure out how to prioritize vulnerability, how to do cloud, how to do software, while attacker extremely enthusiasts about, Hey, let's use this technology, or let's weaponize the model that are trying not to do it. And I think Tropic has published, uh, a recent playbook on how attacker are creating new method and way, and they, of course, they're trying to stop, they're trying to ban them, they go through.

But we start seeing case where LLM are weaponizing vulnerability or are being used to attack ransomware. So AI has lowered the barrier of, of, of access for cybersecurity professional, but also for attacker. And we were overwhelmed before, like I think right now the difference between the dog com bubble and right now is we are seeing this technology put exciting, but we are seeing it as faster growing as a weapon and as any new technology, we are seeing the rush to market.

Of course, Trump insecurity read us as MCP because API security wasn't hard enough, so we needed to have GraphQL. And one of my good friends is saying, I love any GraphQL because I can hack the way through it very easily. And because that wasn't sufficient, we had to create MCP, actually, we released our MCP server and we shut it down for security concern.

I'm proud to say it because we did a threat model on that and saying, that's not good enough. But how many people out there are throwing the MCP out in the ward and saying, yeah, it's secure enough, right? Frank, I I have to pause you for a moment because there are folks listening and watching who know what GraphQL and MCP are good for you, and there's people who don't.

So for the folks who don't give us a little bit of background, talk to me as though I'm my 75-year-old mother-in-law, or my 10-year-old daughter. I, I think you might be right. I, I get over excited about technology sometime and I think that everybody lives in the world of stuff that my brain leaves.

Um, sometimes Only they're really smart ones, All the crazy one. Uh, but thank you for the compliment. I think when we look at the internet, we had the history of API that were soap xml.

So very ancient way to pass data through a system that expose a web interface, and then we kind of settle on rest API that is the standard method where we taught really, really well and long about how to secure those things, how to create that entity. So I think rest API has been around for very long time, but for rest API, you had to create basically endpoint for everything that you want to do. And that is means development.

So some of the dev team has said, why not throw caution out of the wind and open everything to everyone? Just query whatever I want and I expose anything that I want. Because that has worked out well for us in the past.

So that's was the history of GraphQL that you can secure, but it's really difficult to constraint or provide access control because fundamentally you can tell, gimme the information about this, this, and that. And graph will say, gladly, here you go. Uh, do you have the permission to see that stuff?

Hopefully you have your pass through credential or pass through authentication configured. Most of the time you probably don't. So you create, just access your data lake and if you're lucky, you just see what you wanna see.

Um, but it's very difficult to control. Now, MCP have been built in a rush on a protocol that has two or three version, and eight, two A was the evolution of the MCP protocol, but authentication was nowhere to be seen. And all to authentication token being passed through or authentication and authorization have been kind of left in the world.

So we're seeing MCP being exploited up, down left and right because it's a new technology and because it just rely on not very strong foundation of authentication and access control. And that's one of the reason why we shut down ours because our API will build with Phoenix security with specific method in mind. So we put, uh, an MCP server in front of it, and it gives, it gives you access in a different way that we want.

And we expected, so we did a, like any security folk would do a threat modeling exercise. We deem the things not secure enough and we say, you know what? Let's leave the hype to the hype and I'd rather not get hacked than be late for a few weeks.

Um, and that's what we did, but I think we won the few that actually take that hint. That's so interesting. Uh, humans want to use technology to share information and then they end up sharing it with people that they didn't wanna share it with.

And if you're intentional about putting some calls, controls in place, then it takes more time, it takes intentionality. Um, and now we have not only automation, but we have ai. So the problem is just worse, more data, more places for that data to be more places in our supply chain to poison and to steal information from.

And so Frank, I think yeah, please. When you Have, we have, I've been thinking about this very hard and very strong, like why LLM seems so attractive. It's like, why is so easy to get caught into the perception that we have an answer?

And the answer was there is the fact that LLM always give you an answer despite that it's good or wrong or whatever, or whatever position you have, you always could get an answer. It might be wrong, but you always get an answer. So it feels that you making progress despite that you are actually making progress or not.

And that's the intoxicating element of LLM. You don't know anything about API security, I'll ask LLM to do, teach me about API security. You don't have context, you haven't asked us specific things, but it will return you with some stuff.

Um, hey, I have this code. What does this code do? I wanna do this particular things.

It will give you an answer. It's probably wrong. But that's what excitement, because the barrier of acquisition have been lowered, the fact that it doesn't always speed the right information is a different story.

And hence why people that understand how AI was built. I was building bias and network and neuro network back 10 years ago when AI wasn't cool. And me and my co-founder understand really well how AI was built.

And LLM is just a variation of an ai. And if you understand how it works and how to ask the right question, you become a superpower because it really 10 x you. And I think I'm, I'm surprised by the kind of things that if asking the right questions, it will give you the right answer or it will speed up your work, but also can slow you down tremendously.

Or it can create a generation, I think of no brain coder and as an industry, I mean you in threat mode or was we, we, we go long time, me and you, and we're seeing the industry kind of trying to make an effort. I think right now we are creating a generation of people that don't think securely or they don't even understand what a vibe coding. So that's a little bit my fear of creating a generation that doesn't have the understanding or the baseline understanding, but just go with it and vibe with it.

And you can vibe secure coding. I mean, our good friend Jim Monica has created a whole training about vibe coding securely. And I think you can, you just need to know what to do and what to ask and how to ask it.

And you still need the principle to be in there because AI will not magically secure your application. So Frank, uh, what I hear you talking about is a comparison. Uh, there is kind of like the no brain way to use ai, and there is on the flip side, a very powerful way to use ai.

And so my question for you is, what advice do you have for our listeners to be, not the former, but the latter? How can we all learn to be the best users of AI and not the no brain ones? That is a great question.

And for that, we've broken our manifesto. What, what, what, okay. Uh, It's not yet public.

Okay. So we call it, oh my gosh, ai, AI tell us everything. Human first, AI second, human first, the manifesto tell us everything.

So I've been thinking a lot about this, and I think with all this hype, we tend to, we tend to place AI first. You see a lot of company coming out and saying, we are AI first. AI is gonna solve all of the problem in the world, and it's so cool and it's whatever.

Now AI is just a tool and like blockchain was just a tool. Let's try not to create solution before we have problem to solve Engineers. And I know, right?

But in general, if you, if you treat AI or LLM or vibe coding as a technology, as a tool, and you learn how to use it, you become really powerful. And I think in few years, that's what's gonna distinguish the people that talk about by coding LLM, but they don't know how to use it to people that have experience and know when to use surgically technology for that experience. And hence why we say human first, empower by technology like an LM, like a chatbot, like an AI tool to 10 x their capability.

But ultimately you'll never be able to fire an ai. So decision will never be able to be delegated to an agent, but an agent can 10 x your engineers. So if you train your engineer well, junior and senior, to use technology in the proper way, then you have a force of nature.

And I think our attackers have understood that. Well, first, some haven't. Some vibe codes.

Write me the, um, what was it, what's a ransomware letter for the FBI for the director of FBI? Because we have all of that data, uh, and somebody has came up with the same kind of things with Google without any improvement, without proofreading or so on. But in general, you have people that understand this technology and wanna use it and AI second, and you have people that put AI first and they will be left second.

Yeah. And hence the manifesto. You know, I'm so excited for this because it truly is the message the world needs to receive right now.

You know, a year ago, and still today, every board on the planet wants everyone to use AI for everything. You know, every engineering team is being told, use ai, use ai, use ai. No one is talking about how to do it properly, how to do it.

Well, boy, is there a difference between doing a thing Yeah. And doing it Well, I, I can't wait. I can't wait.

Who, who, who, uh, who's coming up with this manifesto? Tell us about the creators. So I generate the first idea.

I sent a few of the leader that you well know, Azar, a few others that have done their first pass on it. Um, few other CISO and, uh, thought leaders as well have contributed on it. We'll have the full, I think we have 25 right now in the us.

We try to make practitioner and CSO alike and non technologists to actually come up with a message that was sustained by both practitioner in security, field leader as CSO in the security field and non practitioner to actually write something. And we wanted to keep it purposely short with 10 commandments that really say, think about these things securely and think about this as a technology. Like that's the underlying mean.

We can go through the manifesto, but that's the underlying message of the manifestos. Like, use tech, use this technology as a technology, use it wisely. Like by code.

Absolutely by code. The hell of things. Um, as A-C-E-O-I push for AI adoption, not AI first, but AI adoption to all my engineering community.

But also we have guard rails and we have methods of embedding things. And we are actively researching how to insert secure prompts in the by coding thing. So they will always return a secure by coded message or prompt.

Like that should be the core of what we do. And the core message of what we do. It shouldn't be, if you don't use a vibe coding tool by Tuesday, you're fired like some CEO have put.

Yeah, I think that's a wrong message because that's, that create that people will adapt, people will adopt, and people will make mistake because it will delegate thinking to the technology. Well, this should be a thinking aid. It shouldn't be an outsourcing and it maybe be unpopular in this opinion, but I'd rather us going forward with the eyes well open rather than creating, what was it, the movie Terminator?

Sorry, I had to throw it in there. You know, Frank, what I like about this is what I'm not hearing from you is I'm not hearing any fear. What I'm hearing actually is a sense of empowerment.

You recognize the power that we have as humans. You know, do we store tremendous amounts of data in our heads? Yeah, we do actually.

You know, do we have decision making? Do we have discretion? Do we have judgment?

Yeah, we, we do actually, you know, and so I'm delighted to hear this sort of elevation appropriately of yeah, the human, uh, and who is in charge, right? The human or the machine. It better how far You can fire a machine.

Like ultimately comes down to that. Like you wouldn't be angry at a machine because it does machine job or it doesn't error or it has a bug. Ultimately, technology is technology.

And we need to recognize this as a technology. That's, we, that's what we, in Phoenix, we created our AI agent as COEs that aid decision making process, but empower people to make those decision. Ultimately, we present three remediation plan.

We don't know better than the engineer. We give you guidance, we give you insight, we give you direction. And we say, based on this, and we explain the reasoning as well, based on this, this is what we do in specific things.

But then if you think that fixing things by a specific asset or fixing things by a specific threats attack vector is better. Choose that remediation method. So we want to empower instead of replace human and security engineers.

I love it. And a lot of people are scared right now because they, yeah, this technology feels like is is AI is gonna replace or go or come for my jobs? If that's the fear, then you're in the wrong job.

You need to elevate yourself to use technology. I think that's where the fear come from. And you have, I think, two sides of people that fear a technology because they feel overwhelmed.

And by all mean this, uh, scary technology because it seems to be able to do everything and nothing. So either you embrace it or you be left behind. And that's the hard truth.

So it's better to embrace it, use it securely, and be at the front edge of this. But if you were doing spreadsheet yesterday, I'm sorry, this will be replaced. Yep.

Hard pill to swallow. Um, but I agree. And uh, Frank, as we're kind of beginning to close up our conversation today, for folks, maybe today's the first time they've learned about Phoenix security.

Tell, tell folks about Phoenix security. So, long story short, we were a bunch of practitioner. They were leading AppSec and cloud set transformation in most of the banking world.

And we wanted to solve a problem that is how do we align executive expectation to engineering action? One of the frustration that we had was when we talked to engineers as security practitioner and our security leader, we tell them, you shall secure your system. And when they look at us and say, what does that mean?

We don't have an answer, or if we have an answer is, well, you need to fix your vulnerability by SLA or you should do threat modeling. Okay, teach me, I dunno, this is a template to use it goodbye. I don't have time.

We don't have scalability. So we wanted to empower, first of all, engineer to understand this is what security expect of you. And then we wanted to align that message with business expectation.

Because if it's not important for your boss as an engineer, you're never gonna be giving attention to a particular problem. So we wanted to solve the problem of security across application security and, uh, cloud security. That is called vulnerability management.

That is a problem that we had for 20 past years, and we wanted to solve it from a business perspective because that's the only way it actually work. And then in that journey, we evolved that with asset inventory. That is also another big problem that we discover in the journey, saying, if we don't know who needs to fix what, how can we tell them to fix stuff?

So we open source our CMDB, yamo based MDB to empower every engineer to declare this is what I own, and I don't have to log into an ancient 1999, uh, black screen with green line system. I can just declare a yamo file apo. And that's automatically confie Phoenix to say, this is the stuff that this team owns.

So if they have vulnerability and you expect them to fix it, we're gonna notify exactly who needs to fix what, where, and ex tell them why it is important. And in a nutshell, that's Phoenix. That sounds really cool.

Frank, if you could go back in time and do the job that you were doing at HSBC, what would it have been like for you if Phoenix Security technology had existed? Well, it's funny that you asked, because that's where Phoenix was born. Incredible.

We created that for ourself in there because we had that frustration because we couldn't translate an executive saying we should do security. An engineer saying, what does that mean? So we created a way for executive to report this is the percentage of security that we want to decrease.

This is the risk level we want to go. This is the amount of money that we wanna reduce in terms of direct and indirect impact. And that very high level message that a non-technical, um, or risk base executive can express, could be translated to engineers saying, this is the vulnerability that you need to fix.

This is where you need to fix. And we as security were coming and saying, look, if you look at this library, these system, these things, you actually maximize your risk reduction. So you will look way better for your boss.

So instead of demonizing engineers who were coming and aiding them to get to their target faster, and look, that was four years ago. So it was a very, um, early stage Phoenix. But that's what the gamification from a business perspective and from an engineer perspective is what have enabled us to move from resolution time of 290 days to 20, 30 days.

Nowadays, it's not sufficient anymore because I think with the latest data that we've seen, expedition time fluctuate between three minutes and seven days, depending on what kind of data source you look. So 30 days is not anymore for critical, but if you don't know who does what, probably you are over a year of remediation. Yep.

Frank, last last thing that I'll invite you to consider doing with me. I want you to teach me how to say your name properly. Can we, can we try this together?

Let's, let's try. Please say it and I'll see if I can repeat. So I usually, it's a funny joke and my partner always makes fun of me because I say I go by Frank for friends.

And then if somebody doesn't call you Frank, it's like, does that mean that they're not your friend? So I don't realize it's, it is something that is ingrained right now with me. But if you wanna try in the Italian way, you did it beautifully actually.

Uh, it's Francesco chip. Francesco chip. That's great.

Yes. Okay. I'm so happy.

Um, gosh. Thank you. Thank you So much.

You honor Italian now. Thank you for your time today. Thank you for your wisdom.

Thank you for the work that you're doing for our industry. I cannot wait to read this manifesto and tell the whole world about it. Thank you.

Brilliant. I think you very man needed. But thank you so much for get having me on this side of the podcast.

It's my pleasure. Folks. Techstrong TV podcast feature your favorite video series, industry thought leadership commentary, analyst research on so many topics including AI and cybersecurity, but also DevOps, cloud Native digital transformation.

Uh, come on over to Techstrong TV podcast to find all of your great content. This has been the AI Security Edge. I'm your host, Caroline Long.

Thanks for being with us today. Our online interactions include audio, video, and sensor data, but most AI applications are still focused on text. This episode of utilizing Tech considers how we can integrate multimodal data with ag agentic applications.

With our conversation with Vish Gupta, founder and CEO of Aperture data, Frederick Van Hern and myself, Steven FoST, welcome to Utilizing Tech, the podcast about emerging technology from Tech Field Day part of the Futurum Group. This brand new season focuses on practical applications of ag agentic, ai, and related innovations in artificial intelligence. I'm your host, Steven Foskett, organizer of the Tech Field Day event series, and host of utilizing Tech Now for nine seasons.

Joining me this week as my co-host is, uh, Frederick Van Herren, who's been present for a lot of those seasons. Welcome Frederick to the show. Yeah, thank you.

Once again. I'm, I'm, I'm mi here as a co-host. So my name is Frederick Van Herren, I'm the founder and CTO of Hyphens, which is a HPC and n AI Consulting and Services Company.

Company. And Frederick and I have been talking about practical applications for AI for a long time. But one of the things that always sticks in my craw, I'm not sure what a craw is, but something sticks there, is that when people talk about ai, they too often focus only on text, basically, it's or bust.

And that's great. And in fact, there's a lot that you can do with text, but text isn't the whole world. Uh, and Frederick, I mean, your background in ai, you, you, you didn't start with text.

No, definitely not. You know, my, as you know, my background is in speech and, and that's the language we use to communicate with people, but we have to understand that people are more visual than learning from text. So a multimodal approach to ai, it's definitely something we're looking forward in this adjunct AI world.

Absolutely. And I think that, um, you know, just regular people who are gonna, uh, be sort of wondering why are we, why are we always talking about, you know, documents and books and web pages and stuff? Why aren't we talking about literally everything we interact with today, which is, um, audio, which is images, which is video documents and so on.

And that is why, uh, we have an exciting guest to kick this season off. Uh, today we have, uh, Vish Gupta, uh, founder and CEO of Aperture Data, who is somebody that I spoke with, uh, earlier this year, and they are really focused on multimodal data. Welcome to the show.

Thank you so much, Steven and Frederick, uh, really happy to be here. So tell us a little bit more about your background and yourself and, uh, and what you're focused on. Um, happy to.

So yeah, as you mentioned, uh, right now I'm co-founder and CEO of Aperture data. Uh, prior to that I was at Intel Labs for over seven years, which is where we started working on this problem. And, uh, you know, just take looking at it, um, from a researcher standpoint and now through the journey at Aperture Data, it's been more like product and user standpoint and businesses standpoint.

Um, for my background, um, I have a PhD in computer science from Georgia Tech and a master from Carnegie Mellon. And I got my undergrad in computer science from in India. So it's been, uh, you know, one part of life where it's a lot of, you know, computer science, deep research, working, really like, you know, uh, on underlying, um, uh, systems hypervisors.

We were one of the first teams to virtualize Nvidia GPUs back when it wasn't even so, uh, popular in, uh, data centers, uh, to, to be, to be offered in cloud environments. Um, and then now this has been a completely different part of the journey, you know, by sometimes, um, mis coding too, uh, but it's a lot more about people than it used to be before. And most importantly, it's a lot about very exciting use cases and applications that have emerged in this last decade.

Just as you know, we have witnessed the progress of machine learning from, just from the very basic, like, you know, the very first image and that thing that make it like, oh, now we can actually automatically understand what's in an image to now AI agents trying to like, you know, order our plane tickets for us to plan this perfect vacation. Well, it does seem though that, um, the progress of ai, you're right, that it started, as Frederick said, it started with, um, with speech processing, uh, for the most part, uh, a lot of the applications early in the utilizing journey. Back when we started this podcast, we talked about, uh, processing images and detecting objects and, and, and processing video and sound and all sorts of other data sources too, um, you know, bio, um, mechanic information from sensors and so on.

But it seems like the prevalence of large language models has just crushed any kind of discussion of anything that's not text. Are you, are you seeing that as well? Um, I think a lot of the practical applications, you know, the, the approach I end up seeing for people a lot of the times is, well, they're asked to go use ai.

So this is like, you know, you kind of have to, you know, we work with a lot of, um, you know, medium to large enterprises, some startups, and we see this very often. There is always this like, how can we use ai if, if they're doing it right, it would be more from the perspective of, there is this business problem, can AI help us here? And then the process.

But then sometimes it's like, we gotta get on the AI bandwagon. There's a lot of funding for it, right? So there is like a spectrum of people.

Uh, but the common thing is like, okay, look, uh, especially when you think about larger companies, data is very siloed, right? Especially if they're collecting, if they're going beyond simple tabular data, going beyond text, that means sometimes it'll be like images, videos, audio, they get organized in places that most of the company doesn't know the, like few people in some team that will have access to it. So just the process of bringing things together and then making sure that the models are up to par, and then can you actually take all of that and combine that into, um, a model that can give you the right answers?

That's a very, um, intensive process and which is what makes it like, okay, well let's prove the value with text first, uh, and then we'll see. Right? Uh, unfortunately though, the problem is in some cases text is sufficient.

A lot of the times people are like, you know, let's say if they are just, um, uh, there is an example where you have a lot of PDFs and granted it took some while to start parsing PDFs to the level of, you know, actually understanding tables and images within PDFs too. It's not, it looks simple, but it's not always, but like, you know, if you're trying to understand a lot of reports that you do internally and enable a rack chat bot, that's the kind of application that you can pretty easily start off with. Um, uh, and, and, and, you know, so you start, if you start seeing the ROI good, but if it is the sort of application where a lot of information was stuck in these other data types and you started just with text, it is quite possible that people arrive at the wrong conclusion that AI doesn't work.

And I've seen that through a lot of, uh, you know, even before we got to the whole rag and stories that we are like Gentech world that we are in today, even when people were just trying to do like, you know, let's say e-commerce personalized recommendation, people wanted to use visual similarity search to recommend products that look like this. Because, you know, we are very visual people, like you said, uh, we're, uh, we look at something and that's what attracts us, not the text description of the product, right? Um, and a lot of the times teams would start building it, but they didn't have tools to be able to query and see what their image data sets look like.

So they would just train the models, you know, with partial knowledge, not do the best job, and then the recommendations weren't as good as, you know, just based on your friend bought this, and so you should buy this sort of stuff. Um, and so the conclusion used to be, well, AI is not helping us. Uh, and, and so that's why I kind of warned in terms of like, you know, it's a text is a good start, but there are a lot of cases where you got to bring in the other signals and it looks very daunting because the tooling also has been pretty broken.

Um, but that's kind of why we are here. Right. So you talked a little bit about, uh, multimodal.

I mean, it's, it's already difficult enough to build models just for text or audio video, um, let alone the different data types, right? Video notoriously are much bigger, binary, very difficult to analyze. Uh, text is easy to read, but much smaller than video.

So how do you deal with all those different data types and those different models into, into one final model, so to speak? Um, I think I, I mean, I couldn't speak, so the way I look multimodal data and what it, everything that it offers, I look, uh, at it at, in, in three, three sections, so to speak. One is the model, what you're referring to.

And you know, there are a lot of vision language models now that are doing really well. I mean, I'm recently reading all about like, um, generative on the SOA side, but also in interpreting like the, uh, if you look at some of the Gemini output and stuff like that, they can really do a very good job understanding what's happening in image video. The second aspect is processing, well, there is no lack of processing today, I might say.

Um, I mean, Nvidia really has changed the name of that game. Uh, and there's a lot of other, uh, inference provider providers and, and, um, some more niche companies, uh, coming up in that space. Uh, and then the third aspect is data management.

And I feel like this is where the biggest gap exists. If, if you look through the evolution of machine learning, you know, we used to see all these papers where, oh, you know, we are now able to detect like this tiniest bit of, uh, like a dog face in the image. And then you went and deployed it in like a medical imaging sort of scenario.

And I literally have an example I ran where the brain lobe, uh, like the brain scan was classified as a telephone lobe. Um, so, you know, so those sort of things, you, they happen because like data has always been the differentiator. The better data you train with, the more representative data you give to your model, the better the outcome is.

Um, and it remains true to this day, but unfortunately, the solutions are still not there yet because changing data systems is a very involved and very complicated process. Um, or it, it doesn't have to be, but that's how we've always seen it. Like, you know, I mean, there's so much, and like people think so much in terms of SQL and relational tables.

Sometimes I run into people, it's like they won't use anything that doesn't support sql, but that's not the right approach. The thing is, what solves your problem if, if AI needs to see all of the data, you need a data foundation that allows you to put all of the data, make it searchable, and make it easy to navigate. That has always been our driving principle behind this, because if you wanna go, um, and you know, earlier we were talking about going from shallow intelligence to deep intelligence.

Um, we are, uh, you know, like I was saying, the models really are very advanced now, the processing power, you know, it's growing constantly and, and, and accomplishing a lot. If we solve the data problem, if we build the right foundation for data layers instead of still cobbling together a bunch of tools that make you inefficient, that, um, create inconsistencies that don't let you scale as much as you can, you're still never gonna fully go into deep intelligence territory. Yeah, it's, it's so true.

And, and what I'm seeing unfortunately is that a lot of ag agentic applications are still focused on structured textual data. In other words, if we're gonna have this process pass data onto this next process, um, in many cases what it's doing is it's devolving, uh, you know, let's say visual data into a, in, in some cases a large set of texts that describes that visual data in either, uh, freeform text or in, you know, structured, um, data and then passes that to the next thing. Um, is it possible for agents to pass the actual image or some abstraction of that image between, uh, agents in these systems?

It, it absolutely is possible, right? So, um, as we were building, so, you know, as you know, the product that, uh, that my company, um, offers is called db. It's this unique vector graph hybrid database that we have purpose built for multimodal ai.

So there was always this aspect of, you know, if someone says, I wanna find images similar to this image, there's of course the vector search angle, so you need embedding generation and things like that, but how are people gonna give you the image? Because they would have to give you the image to put in the embeds, and then you can do vector search, right? They would have to, um, and let's say you were going even further, you wanted to find video clips that had, let's say, kids playing in it.

You would have to be able to understand components of the videos, um, you know, generate embeddings from those and be able to search through them. Uh, but it's not just the vector search part, because what ends up happening is, let's say you want clips where, um, uh, kids are playing in the video, right? Um, maybe the entire video is 30 minute long and there is only like a two minute section in which the kids are playing.

A true multimodal AI database should allow you to search, decode the video and go into those two minute part and just transfer that. Now, why is that important? Because I'm gonna imagine videos, like you mentioned earlier, videos are really large.

Are you gonna be transferring the whole, like, you know, 30 minute long video between different components that need to operate on it? Or do you wanna just take the two minute sections that are relevant and pass those along? So that's like, that there is, there is this whole efficiency angle to it and, you know, not having to wait for hours for something to happen that can only be enabled when you introduce true multimodal understanding in your database.

And that's, that's kind of what we did because, um, you can literally say, I want all the video clips in which a person was smoking or not smoking, uh, and I want them returned to me in thumbnail size. This is one query to aperture db. It does the, decoding it out, the parts that are interesting and it, you know, you know, bundles up the clips and sends them to the, uh, to the next stage.

Um, and you are not duplicating any of this information because remember, you gotta think about scale. We are gonna operate, we, we are operating on petabytes or maybe even terabytes of data, right? Um, and when we represent videos in our database, that is the original video file, but then we very smartly use the graph structure that we have to represent all the regions of interest in it.

It can be interesting frames, it can be interesting clips in it, the same logic with images. You know, sometimes, um, you might, your cameras might be really high resolution and capture a very wide angle, but all you care about is that person that's standing on the street. Why are you transferring all those pixels between the different stages?

Um, so to your, to your original question in terms of why can't you transfer some of these other data? Can you transfer these other data type? I think one is the protocol that allows you to define the stuff, and I think there is still some room to improve.

Like we had to, um, come up with a different query language to support all of this. Uh, we actively chose not to implement into like a SQL or a cipher based query language because they were too restrictive in what we were trying to do. Now we have built plugins to make them compatible because a lot of other tooling lives in that world, but, uh, we started out without hindering ourselves and, uh, we had to introduce the exchange, like, you know, okay, this is how you are gonna give us blobs of various types.

This is how we are gonna demarcate one blob from the other, and the metadata we return is gonna tell you what the rest of it means and things like that. So, um, we are able to do it. Um, and, you know, we, um, work with PDFs, audio images, videos, and, um, it, it, it works great.

And of course, in the backend, uh, we've introduced the whole performance and scale and, you know, understanding that these data types are different. You have, you know, more parallelism requirements, there is less dependency among, you know, individual parts of the data. So there's all that stuff that goes into the architecture to make it high performance and efficient.

And then there is in the protocol to, to enable like to, to define that language. So it's, it's very much possible to do it Right. Yeah, data movement has always been a problem, and as, as, as people collect more data, I think the data movement by itself will get worse and worse.

So how do, how do people interact with your platform? Do you integrate with frameworks or orchestrators, or how do, how do people use and consume the platform? Yeah, I mean, you know, we started out with a database.

No one wants to really think about a database. It needs to be hidden behind stuff. Um, and so yeah, we have, have, um, uh, you know, originally when we started, because we were looking at a lot of, um, training and inference sort of use cases, we integrated with TensorFlow, vertex, ai, these sort of, uh, frameworks.

Um, then we, you know, with the rag in the rag world, basically we introduced Lang Chain LAMA index integrations, and now we are looking, looking into agent memory, uh, frameworks to integrate with them. Um, we also do, you know, so ADB has grown from just being a database into this entire platform where, um, you cannot just manage and search the data. But, you know, we have introduced workflows to make it easy to upload data to, uh, generate embeddings to extract information.

I mean, we have a workflow that, you know, you give a URL and outcomes are rack chat bot, uh, you don't have to worry about segmentation mo like, you know, embedding models and things like that. Um, so, and, and we have these various, like, you know, MCP server plugins, SQL server plugins, so that has grown, uh, now into a platform so people can interact with ADB directly on our, uh, cloud platform. They can use Community Edition, we can do A BPC deployment.

Um, but we also have our own ui. And I'm actually really pleased recently with all the developments that have happened into our UI because, uh, you can literally go to, you know, one of the tabs in the UI type a text question, and if you've like, you know, ingested the data and generated embeddings, it'll show you, okay, these are the images that match, these are the PDFs that match, these are the videos that match and all of this on one interface. And there is still so much to, to, um, improve there.

Yeah. So the, the, the, the workflows or the plugins are, are kind of starter kits and I guess for, for, for the, for the consumers. So can you talk a little bit how the platform works with, uh, enabling autonomous and semi-autonomous AI agents?

Right. So, um, there are different ways that you can go about it. Like, you know, a lot of the agents behind the scenes when they wanna interact with data, they basically might, you know, just do vector search queries and then, you know, implement their own LLM like feed, like do the semantic search and feed things to the LLMs and then generate the responses.

Um, so that's like very fundamental way, which, you know, you can just use the vector search support we have. You can enhance that with graph rag, sort of, you know, like rag improvements to start including the knowledge you've contained in the graph. But where we are seeing this go is essentially introducing this memory interface, uh, because if you look at, you know, the memory frameworks, there are, there's the component that actually takes, um, user log user questions and extract preferences and, and, you know, relevant personas and stuff.

But underneath it ends up storing this in either database or a combination of vector and graph database or just simple text logs. And aper, TVB is perfect for storing all of that stuff. I mean, the throughput and latency we offer in terms of updates and queries, it's phenomenal.

Um, and so it makes a really, you know, good foundation. And so now the, the thing we are working on now is like, okay, what's that memory layer, right? Like, you know, we start by integrate, we, we'll basically integrate with some of the frameworks already out there.

Um, so the agents can really like, make use of the memory and scale, um, through what Aperture DB offers. So I love this talk of, you know, moving beyond text. I mean, that's the, the, the premise here at the beginning.

Um, but, um, I wonder if you could help us with some infra or examples or some ideas about moving beyond video too, because of course, multimodal data, it doesn't just mean video, it means all sorts of data types that are, you know, all varied. So what other data types beyond audio, video, video and obviously text are, uh, people looking at with agent applications, and what are some of the use cases for that? Um, well, you know, I mean, um, Steven, as we talked about, a lot of people are still on text.

We are not even at the audio video stage yet, but, you know, so there is a lot going on with voice. So there's a lot of audio information. I think people have realized there's a lot they can even like, you know, even before getting to voice and, and videos, um, there's a lot going on with PDFs because they, I mean, you know, if you, um, we, we, we all create so many reports, right?

And we like to put tables to summarize, we put charts there, um, we have these pie charts and you know, sometimes we put pictures to show the way, like, you know, how our architecture looks, so there's a lot that goes in and parsing PDFs in itself is, uh, and, and extracting information to start making sense and, you know, at what b um, parts do you, uh, how do you segment it and things like that. All of that stuff involves a lot of, um, work. So I've been seeing, um, like people who have managed to go beyond text, a lot of the times it's like, uh, actually this is the kind of progression.

You start text vector search, right? Then you start realizing, well, you know, your text is giving you some more relationships about things, and so can we connect and start, you know, utilizing the relations around it? So it naturally kind of progresses into well graph sort of notion, can we build a knowledge graph?

Can we use that information to improve the responses? Then it moves into like PDFs and that auto automatically gets into like parsing images and stuff. Um, of course voice AI companies.

I think Fred, you would know a lot more on this one. They are, they are starting to get, uh, you know, voice in my understanding started with like, let's convert this audio into the transcript, and again, go back to the vector search where I think there is an increased understanding around like, Hey, if you did that, you lose the emotions, you lose the, you know, background information. And that's sometimes really important.

Um, so you go beyond that. If you get to videos, there are some cases, especially in, um, medical imaging sort of cases, you know, where the scans, um, like, you know, nowadays a lot of the CT scans or ultrasounds can be pretty like, uh, 3D formats that can be, um, the neural scans are in a different format. Uh, so when you go into more, uh, specific, like more domain specific use cases, then the file formats start to be different.

So that is, um, you know, can you understand, um, the medical imaging file formats and start and, and enable the medical co-pilot sort of use cases, right? Because I mean, patient information is naturally multimodal. Uh, then there is, uh, satellite imaging sort of use cases like, you know, what can you gain?

And that can feed into, you know, traffic sort of things, or it can feed into agriculture sort of things. But, um, something that encapsulates the GIS formats and, you know, understands the different layering, like, uh, a satellite with different resolutions, how do you align pictures from all of those? So there are different formats for that.

And, um, I think there's a lot of development needed on that, on those applications from the model side too. So I think that is that we are gonna see those things come, uh, you know, there'll be more like more dedicated companies first even figuring out the models to operate on these sort of images. I mean, in the past when we looked at medical imaging, um, formats, we essentially would slice them up.

Like, DICOM is like a series of, uh, p and g files, the usual image of format. So we would slice it up because DICOM itself contained too much so that there's like, you know, um, but that's when you become very domain specific. Yeah, I'm glad that you brought up, uh, medical, because I think that that's definitely an area where we're gonna see a lot of development in this, but also as, as you talked about a lot of geographical data, um, I was talking to somebody who's working on drone technology and they are working with everything from, uh, you know, GPS data streams to topographic information like you mentioned to, uh, you know, real time feeds, uh, from sensors.

And all of these things have to be integrated and localized and plotted together. It was a really interesting conversation a little bit beyond me, but, um, but I could understand the challenges because there, you know, it's not just video, it's not just maps, it's not just text, it's all of these things as well as lidar and radar and, you know, cameras and all of this had to be integrated. So I, I think that increasingly that's what, what the challenge is gonna be is how do we integrate all of this data in a way that a, um, an AI agent can understand and act on without just overwhelming it with data?

I, I think you bring up a great point, and, and you know, it, like anything in, in, in AI right now, it's a two part thing. You know, there's the model, it needs to start having an understanding of it. Um, and, you know, there are the, the multimodal models are definitely, uh, you know, um, advancing rapidly, and there is that data part.

So, you know, one of the unique aspects, the why did we bring in a graph into picture originally it wasn't because we were thinking there's gonna be all this knowledge graph use cases and things like that. We brought it in because it gave us a good way to represent relationships, and it was flexible to let us represent whatever data type we wanted to represent in it. So in our same graph structure, and we use a property graph structure for that reason, instead of the, um, RDF, um, graphs that come in that just like, you know, sub we don't do the subject predicate object representation, we do the full, like that if there is a representation for people, it'll be like a person node in the graph, you know, name, last name and all that stuff.

But it'll be, it can be very easily connected to another node that's a picture of that person, or that's connected to like video clips of that person and all of these special data types videos. Um, you know, we can introduce lidars documents, all these things have representation in the graphs. You can go from one type to another, and in the same query you can be saying, I want all of the various data things associated with Stephen and Frederick together, like whatever they appeared together, whether there was a text description, whether there was an event, whether there was, uh, you know, recordings, it, it can go, it can use the power of graph reversal to get there.

Um, so that's why we kind of, you know, originally started with the graph and, you know, now you can basically represent a lot of, uh, application information in it too. Yeah, I think one of the problems too is that, uh, not only is there a large amount of data, but the, the amount of metadata associated with the data is also getting more complex. Right.

So you talked a little bit, a little bit about the medical and geographical, I mean, the amount of metadata surrounding it is, is creating an additional, uh, problem in the complexity of the model. So, so, so one of the questions I, I had for you was, how do you see Agen AI evolved in the next 12 to 18 months? I mean, if you look at MCP servers, they are less than maybe around a year old.

It's going so fast. What's, what's your vision for AgTech AI in the next 12 to 18 months? I think that it's gonna be a lot more focus on what does it mean to get agents in production.

Um, you know, we've built a lot of toy agents, we've built a lot of, uh, like, you know, agents that are starting to do some serious work. Uh, but I think especially in, uh, larger companies, you know, now it's time to go from POCs into production, which means really answer all these questions. So all that we discussed, you know, how does, how does it get the maximum ROI you have to start thinking about your stack.

Like, are you gonna do a framework way? What framework is the best? What sort of models give you the least amount of hallucination and get you the most distance in terms of, uh, you know, your particular use case?

So like, you know, we work a lot in retail and e-commerce, and there is personalized recommendations. Sometimes it, that doesn't require you to be a hundred percent precise. You know, you're recommending product, you're telling them what you can buy.

It's okay if like one of the products you recommended doesn't exactly fall in that umbrella, but we also work with some medical copilot use cases, and there it becomes very important that you do not hallucinate. So the guardrails become really important. So there'll be a lot more increased understanding in terms of, okay, for the vertical that you are in, um, what are you okay accepting and what are you not?

And then what does it mean if you wanna go in production, what are all the data types you're gonna have to involve? What teams have to come together to put this information? What are the guardrails that are gonna be, how are we gonna evaluate?

How are we gonna observe and monitor this stuff? How are we gonna capture user preferences at, at scale without disturbing their experience? Um, I feel like there's gonna be a lot more, uh, you know, focused and organized efforts.

And so the tool like, you know, platforms like ours, uh, become really, uh, key in, in making that happen. Um, I do wish though there is also some effort around taming compute. We've been throwing so much power, and you can see these numbers about, you know, the electricity consumption for AI applications as like literally been drying reservoirs in places because of cooling.

Um, I really hope there is some effort around that too, to reduce the energy consumption. You know, it's interesting. I was just gonna say, it's almost like people need some kind of special database that can handle all this multifold data and maybe a platform that could bring it all together.

Um, yeah, it, it is, uh, I, I think what people need to know is they need to know that such a, that such technology exists and that it is possible to bring together various data types and with AI applications and that, you know, people are working on this because I wonder how many people are just, you know, sort of dismissing it outta hand and saying like, we just can't handle this right, or we don't know how to handle this. Um, so I, I guess, um, what do you see happening next, uh, from the industry overall in terms of integrating multimodal data with, uh, agentic ai? I think it's gonna, I, I think it's gonna increase at a much more rapid pace, uh, with the, I mean, you know, there is at anytime the big, uh, big companies start talking so heavily about it, you know, they start talking like six to nine months early because they're trying to build up hype around it.

But if you, uh, Nvidia, GTC earlier this year, or Google next, or like, you know, reinvent late last year, multimodal was already the thing and agents were already the thing and it was naturally like multimodal AI agents, right? Um, but of course the practicality follows a little bit behind it, all of this. So, um, yeah, so I, I think we'll see a lot faster adoption, especially like, you know, we are in, we are in production, so, you know, people can really unlock the data part, and the moment you unlock data, um, the computer is ready.

Alright, well thank you so much for this. It's been a, it's been a very thought provoking as was, you know, our previous conversation. And I hope that our listeners are starting to say, wait a second, maybe it's not, you know, just about text and just about structured data and, you know, passing JSON between, you know, agents and things like that.

Maybe it's, maybe it's more than that. And hopefully that's the sort of thing that can come from this season of, uh, utilizing tech where we're gonna be talking to a bunch of folks who are doing some really cool things with AI agents. Um, before we go, um, please, uh, let us know where can we connect with you, where can our listeners connect with you?

Where can they learn more and where can they con continue the conversation? Yeah, so I am very active on LinkedIn, so please connect with me on LinkedIn. I suppose you'll share the profile, um, as part of the description.

io or um, docs do aperture data io. Uh, and I would say give it a try. The cloud has free trials, so if you sign up on cloud aperture data io, um, you can try out the database, you can try out our various workflows that make it really easy to ingest existing, you know, data examples, run some embeddings, try out the ui, everything is there.

And if you are concerned about, uh, privacy because you know, you work at a company that won't let you send data to a SA tool, then we also have, um, free community edition on Docker hub, and you can definitely try all the database features, uh, through that as well. And we would really like to grow our community. We have a Slack channel, uh, and we really, you know, uh, amplify people who build and contribute, uh, to the set of applications that can help end users.

Um, so for sure, looking forward to such contributions and more multimodal agents built on top of adb. Yeah, I can't wait to see what people build. Um, and Frederick, how about you?

Yeah, I'm also active on LinkedIn. com websites. And you will see both of us at AI Field Day, uh, which is coming up real soon here at the end of October.

Uh, we're pretty excited to, uh, be bringing together a cool group of companies, uh, talking about various, uh, elements and aspects of ai, some of whom you will hear about on this episode, or this, I'm sorry, on this season of, of utilizing tech and, uh, hopefully some of whom, uh, we will connect with further. Uh, if you are excited about AI and, uh, agentic AI and, and where this is all going, uh, do check out the Tech Field Day website. com.

Uh, that's the website, um, tech Strong AI is our media site. And also, uh, we're gonna be launching another podcast, a weekly podcast focus on AI as well. So keep an eye out for that.

So thank you so much for joining us and listening to this episode of Utilizing Tech. Uh, you'll find this podcast in your favorite podcast application, just search for utilizing tech. You'll also find us on YouTube.

If you enjoyed this discussion, we'd love to hear from you. Please give us a rating. Please give us a review.

Uh, please subscribe. Uh, this podcast is brought to you by Tech Field Day, which is part of the Futurum Group. com, or connect with us on X Twitter, uh, blue sky, mastodon, or, uh, yeah, LinkedIn.

Uh, you can look for utilizing tech. Thanks for listening, and we will see you next week. Okay, the Nobel Committee goes Quantum, you're watching Textron Gang.

Hi everyone, happy Thursday, man. This week's flying by Earth Thursday, Mike, you're still, but you're still in Barcelona, right? I am.

Last time I checked and let me look out the window, but yeah, I am. Alright, it's a week in Barcelona. Good for you.

Um, guys, we've got a, a great show for you out there today on this beautiful Thursday. We are gonna talk a little quantum, a little ai, a little, a little Atlassian, Atlassian. We've got some good people to talk about it with.

Let me quickly introduce you to him. I mentioned Mike is joining us in, uh, Barcelona, Mike Ard, happy with the Yankees win the other night. We've got John Schwartz out in Silicon Valley whose voice is a little sketchy, so we'll try to go easy on him.

And of course, joining us, my friend Jeff Reisch. Jeff, Jeff Re Jeff is the CEO of the IDSA. And always a pleasure to have him on here.

And we found out today Jeff was a quantum, uh, not quantum, a physics major school. Yeah. Larger than quantum.

Yes. Much larger. Billions in billions.

But, um, so how Apro Pro to have you on here, so Mike, the, the, the Nobel Committee awarded the other day, uh, their physics awards to three, uh, physicists who did some really, you know, foundational work in the mid eighties that, uh, kind of forms the backbone, the basis for what we call, you know, quantum computing today. Whi which I think is maybe slightly overdue after all this time, but I guess better late than never. But we haven't seen quantum computing machines in the mainstream.

At least we've seen them being used in various use cases in pharmaceuticals and some other places. It's interesting and compelling. But Alan, you wrote this article on text drawing it and you know, as you were mentioning, it kind of does sort of make quantum feel real.

They got a, they got a Nobel Prize for it, and this is now a, a, a real thing. We are computing with the fundamental elements of nature. So we've been, we've been computing with the fundamental elements of nature all along, haven't we?

Everything comes from nature. But that being said, rumor was they, they were holding off on awarding the Nobel until we did have a, uh, commercial quantum computer, but they were afraid that there would be no one left on the committee. By the time that happened.

They weren't to the, I'm only kidding, they didn't really say that. But yeah, it is a long time calming. This was work done as you know, 40 years ago and hard to believe 40 years ago this was done and we are still waiting for Q Day.

That's the bad news. The good news Q day is closer than ever. So they say, um, you know, we, we have, I I think over the last three to five years we've actually made a lot of, uh, breakthroughs in, in qubits and, you know, all the multi qubit machines and of course post quantum crypto cryptography algorithms are, are pretty, uh, widespread these days.

But, you know, I think number one, honoring these three men who did this foundational work on, which all everything we're doing now is based, is long overdue and congratulations to them. Secondly, I do think it's another drumbeat in the steady drum roll of quantum being real and Q date coming third, I think most of our people out here are not really sure other than post quantum cryptography, right? Breaking down encryption, what are we going to use quantum computers for?

Jeff, if you don't mind? Well, sure, I'd love to expound on this because quantum computers, I think in the long run can be used for everything. But you know, just like every other computer, it's a matter of cost, resources and availability.

But I, I think from a, a security and technology perspective, you're gonna see quantum computing further embrace what we do for connectivity be, because keep in mind, quantum particles can pass through objects there, there, there is no physical barrier to them. Um, now that's an abstract for most people, but we're going to get there with computing, however, and our, our second topic coming up later on with ai, it's directly tied to that because both of them require a whole bunch of cooling and a whole bunch of power of, of electricity. And we haven't solved that problem yet.

I would offer let's target AI and quantum towards solving our energy issues so that we can get more quantum computing and Q day can be sooner. But the sort of things you're gonna see coming out of this, I think when you talk about artificial atoms, I think you're gonna see, um, depending on ethical, how it's looked at by different organizations. I think you're going to see gene therapy come out of this.

I think you're gonna see new, new drugs that can target specific, um, issues and ailments and conditions without having other side effects come out of this. From a technology and computing perspective, I think what you're going to see tied to AI is a faster implementation of whatever it is we need to do with computing. And I know that's a very broad statement, but there really are no bounds.

If you wanna focus on how quickly can we, uh, encrypt something, um, how quickly can we transmit it, how quickly can we calculate what it's gonna be, all of that's gonna come out of it at or around Q day. I also think, uh, quantum entanglements, which aren't necessarily mentioned in the, um, prize notification, at least I didn't see it when I read it, uh, are going to give us a security, uh, condition that we haven't been able to achieve before. And if you don't know what a quantum entanglement is, and this is some of the work they they did 40 years ago, you, you have two elements.

Um, not chemistry elements, but, but two objects. And they are qu there is a quantum entanglement between the two. Anything you do to affect the first will have the same effect on the second.

And anytime you make any change to that entanglement, including simply observing it, the entanglement breaks. Now isn't that a wonderful way to have a secure connection that says anytime it's even observed, it breaks. So thi this is where I, I need, I need a little mind expanding, you know, mushrooms or something to help me out here that the, the, the quantum entanglement thing, right?

Because it could be a particle in another galaxy or another, you know, star system, but yet the, the, the entanglement is instantaneous, but of course when it's observed it, the entanglement is broken. So how if we can't observe it, how do we know it was there? And I know they've gotten experiments that show this, but you know, I stopped smoking dope a long time ago and I just, I can't wrap my head around this.

Well, maybe you stop too soon. That's the problem. There you go.

There You go. But, um, if you go back to Heisenberg, who really started the whole concept of quantum mechanics, the Heisenberg Uncertainty PR principle, really, that's where it came from. There is a cat in a box.

I know I'm going back to, I think many people have heard this sure metaphor, right? There's a cat in the box. There's a cat dead or alive.

Well, you won't know until you open the box. You observe it, therefore you break the entanglement that was there. That said, is it one or the other?

Both conditions were true because you didn't know what it was. And then as soon as you observe it, you know it's broken. You know, which one now exists.

I don't know if that helps. But with the entanglement, the, the good news about quantum entanglements is, as I said, it can, um, they can go through physical barriers. Space and distance means nothing.

So you can have a quantum entanglement with two particles, uh, different sides of the universe. And anything that affects one affects the other. And any time any change occurs to the entanglement, it's done, which means you've opened the box and looked at the cat.

Does that help? Hey, can I mention something that it's, it's, maybe it's a parallel, but what, what I find so interesting about this in this, in terms of what these folks have, these three gentlemen have done in terms of, i, I, I think a, Alan you mentioned the artificial atom, which bridges quantum theory and, and yes. Uh, quantum theory and engineering hardware.

The thing that kind of I see a parallel is a couple weeks ago I went to Stanford. I was at this reception basically about the semiconductor industry and the found the members, the guys who made it all possible. And I think I see this parallel here too.

And I remember at the Stanford ceremony, they were talking about the fair children and what begat the, from the Fair, fair children like Intel, a MD alter, et cetera. And I think this kind of marks that same type of moment where you acknowledge the work of the people who build the construction of the house, the foundation of the house, and makes everything possible. And it seems like this was long overdue, but you know, it's kind of gives me the sense of dejavu about what's happened in terms of the long forgotten people decades from now when AI takes off.

So for those who were, um, under 60 Fair Children refers to Fairchild semiconductors. Yes. Yes, exactly Right.

Yes, yes, yes. Sorry about that. No, I'm dating myself and Mike, I know in all of us.

But I'm thinking about Robert Noyce, Gordon Moore. I mean, these were giants. You all know who they are.

Sure. And I think I wish more people knew who they were. Right?

And I think these three guys, maybe they will be forgotten, but their place in history will be in the firmament forever. Yep. So I have questions.

Am I, do I need all this current AI stuff? If someday I have quantum computing, will quantum supersede ai or will these two things come? I won't.

No, I, I think, I think AI becomes the catalyst for quantum, right? And, and, and so, you know, you've got power squared between the two of them that, you know, because for a, some people say for AI to reach its ultimate goal, and if you believe the ultimate goal is super intelligence or a GI or what have you, you're, you're truly going to need quantum kind of, uh, power quantum kind of computing resources for AI to reach that goal. So they actually are very complimentary and feed off each other.

You know, I, I did two interviews with the, a, a father and daughter, the daughters of PhD outta Stanford, actually her page PhD's in ai, but she's CEO of this quantum computing company called Q Secure, QU Secure. And her dad, Dave, who's also very well versed in, in quantum. And the, the, the thing that blew my mind is when you, when you look at cubits, right?

And, and which is the fundamental bite of the quantum world, of the quantum computing world, right? In conventional bytes, computer bytes, it's zero or one, right? So if you have, let's say, a 64 bit processor, you have 64 different bits, that could be zero or one.

So how many permutations of that can you have? Well, 64 times 64, excuse me, 64 times 1 28. 'cause each one could have two different states, right?

64 to 1 28 a qubit, 64 bit qubit, let's say has more computing permutations. There's more computing power than what we compute in a year in traditional computing, basically, right? Because it is, because it could be both or neither, right?

There's actually three. So that the amount of computing power that a quantum computer brings to it, and I, I didn't do this justice. Go back and look at my interview with Dave.

I think it was a black hat. Um, but the amount of computing power that it brings to bear virtually equals the compute, the entire computing power in the world for a year right now. So, you know, what's that gonna do to ai?

But if I could offer in, in aeronautics is a term all force and no vector. That means you have a whole bunch of power, a rocket or a jet takes off, but you're not steering, you don't have navigation. You're just going, to me, that's quantum computing.

AI adds that guidance and altitude and direction. Excellent. I mean, so here's the bottom line though.

We've mentioned some of the uses, protein folding pharmaceutical, you know, and, and biomed kind of things. Of course, post quantum cryptography and, and you know, cryptographic kind of stuff. Um, modeling, uh, uh, weather modeling.

And, and, and you know, what, what those kinds of things. I mean, just anything where, you know, you, you, you need that massive amount of, of ability and, and, uh, of, you know, of computing power in, in on the head of a pin. You know, quantum does that for you.

However, let me also say, and I, it's in my article that we haven't solved all the issues quite yet, right? Basically for the qubits for quantum computers to be working now, they've gotta be like flawless or something like that. And we are developing what they call fault tolerant qubits, which will be easier to manufacture, easier to maintain, right?

The work these guys did in 1985 is they had to have things in a superconductor kind of environment that damn near absolute zero kelvin's, and there's a locket junction or something where the magic happens. Um, I'm, I'm leaning over my head there, but in any event, right? Fall tolerant cubits make it a little easier for, for quantum computation to be done without all of those things.

There are, Jeff, you mentioned power, the power consumption and cooling and all of that stuff to bring it to absolute zero and all that huge, right? So this isn't the kind of thing you're going to just run because you feel like doing, you know, you want to do some basic trigonometry, right? You could use your slide ruler.

This, this is, this is an expensive proposition. And, and so the, the rewards have to be equally as valuable. Does anybody know?

I mean, maybe Jeff, like, so how exactly do I write software for this thing? I mean, is there isn't a traditional compiler, right? So there's gotta be something that I'm using to in access those cubits that Alan's talking about.

But what is that? So first of all, it's, it doesn't really exist yet. So it's hard to say what's actually gonna work.

But I believe what you're gonna see is rather than, I think compiler can be a very quaint term when we get to quantum computing, because you're going to, I believe what you're gonna do. And once again, I think AI and quantum are tightly coupled. You are gonna be able to take a concept and say, here's what I want to be able to create.

I believe AI can create the programming that you need to run on quantum. In fact, I don't think right now a human can comprehend writing a program that can truly take advantage of that advanced speed. That's why I think AI is gonna have to be the very next step, if not permanent step to doing that.

Wow. Alright. So you're telling me the universe is right here on my thumbnail, perhaps, right?

A little, a little animal house there. Animal House. Yes.

Yeah. All right. Hey, we're a little overtime though.

We need to take a break. Let's come back. And again, we're gonna talk a little AI now.

Uh, I, big news from, uh, IBM, you're watching Textron gig. You've earned it. The spotlight, the responsibility, the weight of teams, companies, and entire industries fall on your shoulders.

Lives depend on your decisions. Your home life included that work. You are protected physically and digitally.

Nothing gets through your team without a fight. But in a globally connected world, everyone sees you, including those who mean to cause you and your organization harm. And now home your sanctuary attackers see an opportunity.

Your digital front door is wide open. And what compromises your home can breach your boardroom. Because the devil's greatest trick isn't targeting your workplace firewall.

It's convincing you that your personal life isn't at risk. Black clerk, digital executive protection, defending the new attack surface your personal life. Hey folks, we're back.

And yes, as Alan alluded, we're gonna have a little chat about AI this time. It's IBM making some news with a partnership with Anthropic, and they're gonna integrate the clawed tools into their software development projects, gonna show up in mainframes, all kinds of fun stuff that John covered. But John, um, do you get any sense of irony here that, you know, the folks who gave us Watson are now reaching out to get some help from Yeah, I know, I know that, did you know that that's a pretty big, uh, thing there, symbolism, right?

I mean, exactly. IBM claimed that they were the beginning, the forefronts, the builders of AI in a sense that they brought it to commercial use in a sense, they're going to anthropic and Anthropic is gladly gonna help 'em. Uh, so yes, it's an interesting deal in a, in a sense that IBM greatly expands its AI capabilities through a startup.

And it also for Anthropic, I think it's incredibly significant and plays into this narrative that Anthropic has been pushing very hard lately to get into enterprises. Um, there are a lot of companies trying to get into the enterprise through AI capabilities and with large language models like Claw becoming more central to business software platforms. In a sense.

Anthropics been making this push into the market of enterprises, um, as kind of the go-to vendor for large organizations. In fact, they've been courting corporate clients since they launched Claude Enterprise in September. They claim they have 300,000 business customers.

And one of the biggest, I think the largest enterprise agreement to date was announced Monday of Philanthropic. And it's with Deloitte's. So that would bring Claude AI models to nearly half a million employees at Deloitte.

Um, it's also on top of that, anthropic has for ties with Databricks and a partnership they announced, I believe in March, which is targeting businesses looking to develop their own AI agents. You know, this is kind of this divergence between what Anthropic is trying to do. They claim they are now like the enterprise go-to solution versus OpenAI, which is, uh, stressing the consumer side.

But, um, it is interesting that IBM chose to work with them, given the history of Watson and whatnot. And, um, I think we're gonna see more of these types of deals between the legacy and between the startups. Uh, it's just really gonna be, uh, almost like a weekly occurrence, I think.

Isn't it refreshing to see an AI deal where there's not a couple of hundred billion dollars thrown around? Remember? So like, Hey, I, I've got my shimmy says this afternoon at two 30, uh, and it, and it, and, uh, it's Shimmy says, uh, when the bubble bursts set to the, to the tune of LED zeppelins when the levy breaks.

And, um, and there's also a companion article to this up on Techstrong it that I'll send you to as well. Great. Everybody, you know, to me, this sounded a little Barney ish.

I don't know if the guys were in purple when they were up on the stage. You know, we're all working together. It, it is ironic that the folks who brought Watson, you know, the first kind of AI we thought about are now, now, but, but IBM's like that, you know, they'll, they'll put anything in their chin.

They got the pipe. If you give them product to put in their pipe, they put it in their pipe. Um, but here's the real question, guys, and I don't want to, you know, stay tuned for my shimmy says this afternoon, but when half of the growth in the US GDP for the year is due to data centers and AI deals, half of the, there's more money being spent on data center and AI deals this year in the US than consumer spending.

Consumer spending is usually 70% of our GDP, but there's more of that being spent on data centers and ais as, but here, the, the, the downside is, if you look at the amount of money we've spent on data centers and ais, it roughly equals the entire GDP of Singapore, which is a pretty sizable GDPI think it's 500 billion plus dollars or something like that. The amount of revenue that AI has generated roughly equals Somalia, okay? Single port of Somalia.

com bubble burst, and they haven't lived through the real estate market crashes. And they maybe have been through the great recession, though we, in tech, we kind of skated at, right? This is a bubble.

And everybody's jumping on making Barney announcements and, and pledging hundreds of billions of dollars that they don't have because they're getting hundreds of billions of dollars from another company who doesn't have it. And that company gives it to this company, and this company gives it to that company. It's Enron all over again.

Yeah. You feel like burst you part of Go ahead. It's gonna be in, it's gonna be interesting.

We hear these projections of what they're gonna spend on data centers, and it's like more than a trillion dollars over the next four years. And when the reality rubber hits the roads, when we finally get to that point, we're gonna find that it's a fraction of what they said, because it was impossible. These numbers were just being thrown out there Randomly.

Well, to Jeff's point, you don't have enough power to run 'em or enough cooling to water to cool 'em and come, and that's assuming that we're really going to use it 'cause it's really gonna generate money. But he, here's an interesting thing, and again, stay tuned to my shimmy says, for this past bubbles were funded by debt, brought down the savings, and, you know, the savings and loan crisis, right? com, it was all VC money.

This bubble here, it was primarily, or at least initially funded by CapEx from the large tech companies who were sitting on billions and billions of dollars. And now they spent it. But now the second generation of these AI companies, you know, the neo clouds and, and so forth, their, their debt, their, they're funding with debt.

And, and when this thing, you know, when the music stops, those people who don't have chairs, it's gonna be ugly. Hey, John, are you reminded of the last time IBM partnered this deeply with a small startup company and how that all turned out? So, uh, so I'm like, Yes, I know, I know you Kelly.

You know, can I, can I just go back to something that a said, then I'll let Jeff, I'm sorry, Jeff, but it's, uh, in Silicon Valley, the big question is, is bubble or you know, it's bubble or boom, right? Well, they're at war people now anticipating the bubble to burst. Right?

Now, what they're trying to figure out is where are we in terms of that happening? Are we at 1997 point, or are we at the 2000 point, or are we getting even closer? And a, there's several people I've been asking this of, and they're convinced that within 18 months we're gonna find out and certainty when that happens.

Nobody knows, of course, but it's inevitable, as Shimmy says. I, I would, I would agree with that. I, I do think that we are, um, it's near, it's near to medium term.

It's not a long term issue. Before this bubble bursts for a, a number of reasons. Real estate power, environmental, downstream issues, finding enough talent to be able to manage all this.

And then getting to the point of now that we're there, what do we do? Because it's, we're, we're, it's all forced, no vector. I'm going to use that one again.

We're streaming towards this, and we don't necessarily have an end goal Right now. The goal is how much money can investors make? That's, that's what this is all about.

And when that, And it's fueling the whole stock market Yep. And when that breaks, it'll probably break pretty big. It's gonna be catastrophic and, and the thing about it, but here's the real piece that's missing.

Where's the killer app? What is going to generate the revenue to justify this kind of spend, right? I need 10 trillion in revenue to justify a trillion spend.

And we're, you know, we, we we're not sure it really works at that level. It's great for writing, you know, helping you write, it's great for some marketing stuff. It, it's getting better writing code, but geez, geez, I don't know.

Yeah. There's not a lot of logic. There's not a lot of logic applied to this.

It's all pipe dream. And, and, but it's funny, Mike, you mentioning IB I'm working with startups. When you think about Microsoft, I think about even Apple, remember they had that partnership with Apple mm-hmm.

With kaleida intelligent, whatever the hell that thing was called. And it, these, these things kind of went awry for fraud. You know, the AI train is speeding down the track, and companies are afraid of being left behind when it leaves 'em fomo, but they don't know where it's going to end.

Is it gonna end at a station or is it gonna end at a, in a mountain at, you know, without a tunnel? No, It's like that meme, I'm sorry, go ahead, Mike. If you look at the open AI numbers, it would suggest that they are also supplementing the cost of everybody's prompts.

So every time you put in a prompt, you know, if they charge you a buck, it's costing them four bucks and they're not quite Figuring it out, they'll make it up in volume. Sure they will. com days, right?

I how many times I heard that in the dot coms, we'll make it up in volume. It costs, you know, I, I charge 90 cents. It cost me a dollar, but I'll make it up in volume, You know?

Well, you hit, you hit the nail on the head when you mentioned Enron. I've been thinking about that more and more recently when I say this, this doesn't, it, it just, it's just like froth. It's hype.

It's over expectation. It's b******t. And it's, it's gonna come back to haunt a lot of these guys.

And it's, it's an, it's inevitable. That's all I really could say about it. I, I wanna put a, a positive spin on this if I can, because it is real and it will benefit us.

Yeah, that's true. And it will happen. It's the investment part of it and how much leverage our, our economy has on it, versus is this ever gonna really happen because it will.

Yeah. No, look, you know what, Jeff, it took about 10 years after 2000, but we did use up all that dark fiber in all those data centers we built. Right?

And we need new ones now. So AI is real. But, you know, thi this is a classic bubble scenario.

And, and what's even worse is you've got the US government, you know, we're already 35 trillion or whatever it is in debt. You know, our administration's all in on this taking stakes in these companies when no, when those stakes go south, someone's going to, someone's gonna have to answer for santino, Carlos. And, you know, I wanna make sure I understand what you're saying.

You're saying that, you know, because we have AI and we could all now write better emails that we're not gonna move the GDP needle. Right? Right.

Basically. Yeah. You know, pretty much, I mean, here's a flip side.

How bad would things be if we didn't have AI and data centers to spend money and, and juice things up around here? You know, there was, there was an article in the Times the other day that we really have two economies right now. We have an AI economy, which is booming right on the, on this kind of irrational exuberance.

And then we have the rest of the economy, which quite frankly isn't doing so good. Right? And you know, the, this, this is, it's, it's a scary proposition.

It's a scary proposition. And here, here's another thing, and I've said this before, when I hear people start telling me it's a new paradigm. The old rules don't apply anymore.

Things are different this time, man. That's, that's when I keep my hands over my pockets because you're not getting none of my money. Um, that it, you know, fools, fools in their money anyway.

Good. Yes. The irrational ex, but this is true.

This is irrational. You know what it was? I think I was younger and dumber back in 2000, right?

And I was like, oh, Greenspan, he's putting his foot on this. He's killing it. I'm, I'm, I'm about to retire.

G*******t. But you know, in retrospect, he probably wasn't wrong then. And we're probably not wrong now.

Yeah. By the way, I was in that same boat with you. Yeah.

Then, um, but the second thing is, I, I teach part-time. I've been teaching for a long time. 90% of what I teach hasn't changed in 30 or 40 years.

Yeah. So for every new thing we get, we still have to go back to what are the fundamental principles and do they apply or not? And yes, they do.

Yeah. Agreed. All right.

Hey, I think we're outta time on this segment. Let's come back and we're going to get a report from Mike about what he did on his Barcelona vacation. You're watching Textron Gang, 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 Techstrong Group.

All right, folks. As Alan was saying, I am here in Barcelona at the Atlassian Europe 2025 conference. They're talking up their robo AI agents that are gonna be embedded into every piece of software they have, and they're only gonna charge five bucks per user.

So it's gonna be interesting to see how all that comes together, given our last conversation about the cost and the pricing. But, um, what they are saying is that they're gonna use this MCP protocol that Anthropic developed, along with a graph to integrate all their AI agents. And every, these AI agents will all know about each other, and they will suggest and help us all do various tasks across their entire portfolio of collaboration.

Software includes project management software that people use, JIRA, Trello. We are big users of Trello here at, uh, tech Strong Group, and it's gonna change the way we work. But it was interesting with what they were saying is don't go after big projects.

They were saying a lot of the issues we're having with AI as, especially as noted by those folks at MIT, is that a lot of companies went out and tried to hit a home run with ai. They're saying play small ball. And they're saying automate existing workflows, just move them forward and, and get some AI muscle memory going and understanding of how to use this stuff and when it works and doesn't work.

And that's gonna be the path to success, which frankly struck me as, you know, common sense. But maybe that is the way to go with all this stuff, is just give it to everybody and they will find ways to use it. And then we'll figure out what the big ROI is later.

Alan, what do you say? You know, they'll make it up in volume. Look, so agen, ai, AI agents, I think are gonna be the make or break of this generation of ai, right?

I, I think generative AI has its users, and we've explored them, and we're going to continue to explore them. But when we look at a, you know, agentic ai, these agents that are gonna, like, let's say cello make using our cellos easier and stuff like that, or workflows that will help us, help us publish articles, you know, on our sites. Mike, it sounds great in practice, but as you know, we haven't been able to get it done in reality.

And, and that's where rubber's gonna meet the road. I mean, and now I, I think this is a classic case of Atlassian, you know, keeping expectations low, expect saying, keep it simple. But it also reminds me when DevOps first started, right?

Enterprises didn't adopt DevOps. People adopted DevOps, small teams adopted DevOps, right? For a long time you had the, uh, Spotify squad model, the Amazon two pizza model, right?

And so, you know, it was good for a team of 10 to 12 people. One, you know, there was a huge jump What a similar message as the CEO of the browser companies here. And that deal hasn't officially closed between Atlassian and them.

But he was saying the same thing about this forthcoming AI browser of theirs. Well, it's forthcoming in the sense that, well, it works on the Mac today and it'll be on Windows sometime next year. But, um, it was the same thing.

It was, he was saying onesie, twosies, people will adopt this to get rid of their tabs on their browsers and have a more, uh, integrated workflow experience. But he wasn't expecting big companies to kind of sign big deals as much as it was gonna be led from the bottom up. And people are just gonna experiment with stuff, stuff and move the ball forward.

I think that too, is probably the way to go. And I, I, I think maybe we're just trying to come up with some sort of prescriptive approach to ai, and we should just let a thousand flowers bloom and see what happens Now, who's over 60? I know.

Um, you know, it's interesting. Yeah, this is kind of the same thing that, uh, Atlassian did in, in Anaheim. I remember going there several months ago.

Uh, and it's just a very more practical, reasonable approach. And I think it's, it's the right approach. And I actually think that they, they're onto something.

And, uh, rather than shoot for the sky, let's just see what happens, right? Let's see what happens from the rank and file and see where it grows. And, you know, it is that, that whole mi again, by the way, we, everyone is referencing that MIT report, it's like just, it's getting torched by everyone, but who, who, you know, who kind of took it, took it literally for what it was.

I mean, but if you read between the lines, it was not no immediate effective, uh, profitability instantly. And it was, you know, this is gonna take time and agents, maybe this is the year of ai, AI agents. I kept, I keep hearing that.

Now I'm hearing some people say, well, maybe next year, um, we'll see, We shall see. I'm like, I just don't really see yet. Like, all right, let's say we do have all these AI agents, and I've got 20 and you've got 10, and the other guy's got 50, and we're all gonna put these AI agents to work, and somehow or other, they're all gonna find each other and negotiate with each other, and there'll be some sort of handoff and something magical will happen and it will all get orchestrated.

Uh, maybe, but I don't see that happening in the next six months. The crazy thing is like a company, like, I won't mention a company by name, but there are several candidates then announce, uh, AI agent product only to then change and say, well, in a couple more months, here we have a new version. And this must to be driving people on it.

Crazy. I mean, how do you keep stay ahead of this, of all these fastballs coming at you, these curve balls in many, in many cases, right? And, uh, you know, and it's just, it's, it's, it doesn't, it just doesn't make sense.

It's just an, again, it is baked into the hype, Especially if you've got a CEO who's standing around telling everybody that AI is going to change the entire economy and the world that we were living and working, and the poor it guys probably looking at 'em going, uh, I don't know what magazines you're reading, but not in my world. Uh, you know, I think AI will change your economy as, as Alan referenced earlier, I'm not certain in a positive way in the short term, um, because of the exuberance. Uh, but the, um, but John, to, to your, to your point.

So let's go ahead and say we're all kind of admitting our ages. And by the way, I think I win that contest. But, um, the days of how does it keep up with more fastballs coming at them?

They have to, the days of No, no, no, slow down. I can't handle that much. Doesn't work anymore.

Um, you know, the days of maintenance did, who heard of a maintenance window anymore? When's the last time you heard that? But, but that was the law, you know, 20 years ago.

Yeah, That Was, so I think speed and volume are going to increase. They're here to stay. So we need to adapt.

And we, and, and it could be that we use Ag Agentic IA to help ourselves to do that. Yeah, no, it's just, it's, I guess, you know, we, I I'll age you, I'll date myself again. When you're talking about Moore's Law, that's like blowing outta the water with this, with the, the current, the current environment.

Everything is based on speed, adapt, or die, more so than ever before. I think that's what terrifies people. The, the fact that they're, they're trying to get ahead of something that they can never get ahead of.

They're always playing catch up. It's interesting stuff. Interesting stuff.

Anyway, Mike, thanks for the report from, uh, Barcelona. How are the topics? Is the paella, The food is lovely, as always.

The weather's been great. And you know, I, I could live here, You and me both. Mm-hmm.

Okay. Adios from Barcelona, Jeff, John, thank you for joining us. Thank you for joining us today.

As usual, we have full text on TV immediately following our gang episode. And a reminder, again, two 30 live on X and on LinkedIn, I'll be doing my shimmy says we'll dive into this AI bubble business, and, um, got a lot more to do. And of course, we'll be back tomorrow with our week and our week wrap up Friday show for Textron Gang.

Until then, though, on behalf of John, Mike, Jeff, and all of the gang folks here at Textron, have a great day, everyone. We're out. Hey everyone, it's Alan Shumlin.

We're back here and we're back live at Swamp Up in the beautiful Napa Valley. You couldn't ask for a better location. The sun has come out, you know, it's good for the grapes.

They say it gets a little cooler, a little warmer, the sun, the rain, you get good grapes, good wine. But we're here with really maybe the highlight of our panel today. Um, three, three of the VIPs of, of the, uh, event The man to my immediate le Actually, I'm gonna let you go last, okay.

Let me start to my far left, I wanna introduce you to Rahul Tripti. Rahul is the GVP and GM of the ITSM, something I'm a little familiar with business unit at ServiceNow. Rahul, welcome to Techstrong tv.

Thank you. Thanks for having me here. Give our audience needs, no introduction to ServiceNow, but give them a little bit of your background maybe and a little bit of what you're doing at ServiceNow.

Yeah, so I'm relatively new to ServiceNow. I joined a little over four months back, oh, uh, new. And I'm running ITSM, which is the bread and butter, the largest business ServiceNow does.

That's where ServiceNow was founded. My background has been building products for a long time for large enterprise companies, but the interesting bit is I switched midway to being a practitioner myself. So I was running DevOps teams in the cloud space in my last startup before I came to ServiceNow.

So I've been on both sides of the equation, building products and consuming products. That's, and that, that's missing in too many of our vendors. As someone who speaks to vendors all the time, you, it's good to have that practitioner side to see what it is they, they're actually feeling.

As that goes by, let's introduce Justin Boitano. Justin is VP of Enterprise AI at Nvidia. Justin, welcome.

Thanks for being on Textron tv. Thank you for having us today. Give us a little, maybe a little bit of your background.

Yeah, Well, I've, I've been in Nvidia now, actually for 13 years, I guess. Wow. A little bit of everything over that time, but, uh, you've Seen it come go, huh?

It's been, it's been, you know, quite a fun ride, uh, you know, watching us really reinvent how computing is done. Um, you know, from really the ground up. Uh, and I think it was, uh, when I first joined in 2008, we had just invented Cuda.

Nobody knew what it was. We were going around library by library, trying to port applications onto GPUs. People, you know, thought we were crazy, I guess in the early days.

But eventually, you know, every, uh, overnight success is, is 10 years in the making, right? And so we've been working, It's exactly the biggest secret in tech, the 10 year overnight success. Yeah.

You know, we had Aon earlier, and we were talking about so many people think of Nvidia and they think GPUs and the hardware, but the real secret sauce here is Cuda and the software and the ecosystem with partners like ServiceNow and Jfr, and, and that we had Sonar CEO here as well. Um, and that's really the, the key that's driving all of this is as much as the GPUs do Agreed To my immediate left, immediate left, this man needs no introduction to our audience either. I've had the pleasure of interviewing him for, um, 10 years, 12 years, something long time.

He's the CEO co-founder of Jfr Shlomi, Ben Hay Shlomi, welcome. Pleasure being here again. Again, you, Thank you very much for having me.

So, look, I was in the keynotes this morning. It was an amazing keynote. And as one would expect this year in tech, AI was front and center.

We've been talking about it all day here. The thing about this is, look, all of us have been around the block. We've seen technology waves calm and go flow and flow up and down.

We've never seen something this disruptive across the entire breath of, of our industry, right? I, I think Satya Nadella maybe said it best, or the best that I've seen in that we are moving from becoming, you know how Mark Andreessen said every company's a software company. Yeah.

We, were all software companies, but we're moving from software companies to intelligence engines, right? And what, that's a profound change in our business. Show me if it's okay, I'm gonna ask you to kick it off.

What does that intelligence engine bring that to some of what we talked about today here at Swamp Up ag, agentic, ai, the whole ecosystem, dev gov, ops, all of it kick us off. So, Ellen, I, I, I think we will need two days to cover this, uh, And that some, But, uh, but I, I will touch the immediate things that we see in the market. And this is a, across the board, it goes beyond sectors.

It goes beyond, um, company size. It goes beyond, uh, geographies. What we see is a revolution at this disruption, uh, that changes everything we knew about the day-to-day practice.

And a company like Jfr, we are not the native AI company. We are the infrastructure company. We are the providers of the peak and shovels.

We are not the gold miners, and therefore, we have the privilege to see from below the changes that are happening. So one thing that we see happening across, uh, our portfolio is that consolidation happens not only in terms of technology, but also the inner organization, the CIO and the ciso, the compliance managers, the audit managers, all of them must collaborate in order to overcome the chasm, to bridge it, and to eliminate silos. Otherwise, they will stay behind.

The second thing that we see is that developers, it used to build called, it used to be called, used to deliver software. And then few years ago, they started to be security expert. And now they have to be release expert, and they also have to be AI experts.

0. They are changing the world for everyone. And the last thing that, uh, we see is that if we are not looking at the, uh, opportunity as coexisting, uh, providers, one of us will stay behind.

If we go together, we will change the world together. This is not a race. And, uh, and therefore I'm privileged, I'm honored to walk with companies like ServiceNow and Nvidia, um, to give my customers a better experience, an experience that they expect a one platform experience.

Absolutely. Rahul, I'd like to come to you, right? So you think ITSM, is there any sector of our tech world that is more rule bound that you would think needs a little shaking up maybe, or, or maybe doesn't need a shaking up, but is certainly getting a shaking up with ai?

If you don't mind, share with our audience a little bit of the profound impact that AI is having in the world of ITSM. Yeah. So I think it's getting shaken up.

And honestly, we, being the leaders in that segment, we would like it to shake up because if, the way I look at it, is it of yesterday, yes. Have changed. Now it is true truly a broker of services, right?

They're managing SaaS assets, they're managing cloud assets. So go as the IT of yesterday, the service has changed from IT providing services to I want my self service, right? So when Jensen was on stage on knowledge, he's like, his main thing thing was, I want my service now, right?

That was his mantra. So people want their service now and management is no longer like, top down, let me tell you what to do, what not to do. People are not used to that kind of thing.

So we are kind of reinventing it. Service management and AI actually helps you really create that agility on top of the more fixed workflows, because now you can do more with AI to create value out of the workflows. So workflows can still exist.

Like the example we gave this morning, compliance and regulation is still required because who wants to have an application that is gonna be breached tomorrow? Right? That's at the same time.

Does it take, should it take a week to get approval? No. Right.

So I think we are reinventing those ITSM processes and kind of integration with jfr, looking at the AI patterns and everything else, because the speed has gone whatever, 10 XA hundred x, right? So things that were taking weeks are taking seconds. So that's where our head is at from an ITSM perspective.

Absolutely. It's velocity. It's velocity.

Yeah. You know, talking about the profound change, if we, if I asked a hundred people watching this live right now, what is the AI company? 98 of them are gonna tell us Nvidia, but Justin, you know, this, and even Nvidia, I want To know who all the other two, Who the other two, they, they're living somewhere who knows on a, on an island somewhere talking to a volleyball.

But, but Justin, even Nvidia knows that as great as Nvidia is and is, they've led the charge help lead the charge here. You can't do it alone. You need partners like Jfr, like ServiceNow, like sonar, like, you know, so many.

You, I mean, one of the, the real strengths here is NVIDIA's ecosystem. Talk to our audience a little bit about that. Yeah, I mean, I, I think that's, uh, well, a, a great point.

Um, you know, Nvidia forever, honestly, has been an ecosystem led company. And I think honestly, it's, it, it comes top down at Nvidia. Like Jensen realizes the power of an ecosystem for selling our physical hardware through OEMs and ODMs globally, uh, through infrastructure companies, uh, you know, uh, plumbing, the runtimes of these accelerators, uh, up to the AIOps applications and into the GSIs.

So we work across the entire ecosystem to try and provide acceleration to, uh, I'll call it, uh, key, you know, workloads that we know are gonna deliver a lot of productivity or performance gains or, um, you know, really kind of transform the, the business, uh, if you would. Right? Um, and, uh, you know, this, this latest version of it, I mean, for a long time we did it in high performance computing to, to, uh, deal with F-E-F-E-A and, uh, you know, CAE and like the simulation, uh, of the physical world.

0 where it's the, the reality is it's a probably a $3 trillion industry, you know, a a trillion dollars in, uh, pure software that gets bought per year across enterprises and $2 trillion in services. That entire industry is being rethought now through this, uh, this new way of building software where you've got age agentic systems that can break down problems, try and solve the problems on their own, and then reflect on the answers. And so there's, there's a, a tremendous opportunity, I'd say, for all vendors in this space, really to, to ride that wave with us, uh, in the era of ai.

Agreed. If I may add, add to it, uh, Alan, look at, uh, what Nvidia did, um, in, in the history of software, the first thing that they act was a community native company. They released their software as open source.

Yeah. Um, today, um, uh, Justin and his team presented on stage, like everything they build in order to optimize GPU with software is open source available. That's right.

For the community. Open source is not only we build it for you, but we build it with you, with, with the community. So I, I think it's really speaks for itself.

Uh, absolutely. We are looking at the improvement that software brings to the world of ai. Yeah.

com today, my mom rest us all used to always tell me, show me your friends, I'll show you who you are, right? You've probably all heard that, or similar thing from your moms, I'm gonna ask, you already said it, Justin, but Rahul, and then I'll come to you. Shlomi.

What's the importance of your partner ecosystem in this brave new world of ai? So, I think any, anyway, at the core of it, if you look at ServiceNow, it is only about workflows data. So that's what we have built our business on, right?

Because enterprise has front office data, back office data, asset data. And if you don't unify the data, then you can be disparate systems on top of it. You tie it with a workflow.

So now the third leg of this tool is AI for us, because AI helps you do your workflows better and faster, and there is no way we can own all the pieces of the workflow anyway, right? There is the software supply chain that multiple players, including jfr, are there from a hardware perspective or from a software infrastructure perspective. Then we have cloud vendors, there are model vendors.

So at the very heritage of the company, we believe that partner ecosystem is critical to it, because that's what our customers want. They cannot rely on a platform that is closed, monolithic does not allow for partnerships. So I think it's the DNA of the company.

That's why we are so excited to partner with. We call it like any model, any industry, any infrastructure is what our ethos is from. Excellent.

Yeah. Shlomi, I, I believe that, uh, um, what our customers are telling us is the, uh, the honest truth and the pain that they experience, and they also know how to put the volume on this pain. Is it a major pain or something that we can handle?

And, uh, every time that, uh, that, that technology companies coming with a piece of innovation. The second question should be, what's my ecosystem? And, uh, some people immediately ask the opposite question of asking, am I overlapping?

Am I competing that we are running a platform? We have, I dunno, thousands and thousands of thousands of logos in, in our joint portfolio. For sure, there will be some overlap, but if the one plus one equals more than two and our customers, um, actually ask for it, how can you go wrong?

And when we present presented apras, the, um, the, um, dev gov ops solution we discussed today, we didn't present it as an ecosystem tool yet. It was just an idea in the beginning of the year. And our enterprise customers stopped us right there.

We said, listen, the people who need to use UPT trusts the application owner. They're not coming to jfr, they're going to ServiceNow. And, uh, when we started to, to work with Rahul team, um, and we spoke with the customers, they actually echoed that.

And, uh, and what we've built together and presented on stage here today is just a representation of our, uh, of our customers voice. Uh, I'm, I'm very proud to be part of a company that instead of coming as an arrogant vendor, telling them what is right for them is asking the, the customers what will be better? How can I make your life better?

I love it, guys. I gotta bring up a difficult one. It's not on our list, but I've gotta ask you, I've talked to a lot of people about ai, as you would imagine, it's almost impossible to live up to the hype.

The hype, the hype cycle is there, right? And there are a lot of people out here who are starting to, you know, the ankle biters. It's not everything we thought it was gonna be.

It's not. This is a lot harder than we thought. It's gonna take longer than we thought.

I think 'cause the expect we set such expectations of it changing the world so quickly, and I said this, even if we stopped developing AI right now, and we just said, okay, let's digest what we have, it would take seven years to fully integrate it into all of our ecosystems. But what do you say to the naysayers who are saying, we're not going fast enough. It's not good enough yet we may never get to the holy land to the promised land.

Justin, you're, you're Nvidia, I'm laughing and you're gonna go walk A day in my shoes. It's moving really quickly. Yep.

Is all I can say. And I think, you know, in the, the early days of generative ai, uh, you know, people were kind of dabbling. They, they treated it like Java.

It was a new technology they wanted to upskill themselves, but they didn't know how to apply it to their most pressing business problems. Um, you know, and, and we see now every enterprise really focusing on like, how do I reinvent the core of my business? Honestly, at Nvidia, we've done it ourselves too.

Like Jensen gave us the challenge, you know, double the number of chips that you produce, uh, every other year. So instead of doing a chip every 18 months, do it every year with a design cycle in between. And the only way to innovate at that pace without obviously exploding your workforce, is by using AI and infusing it into the, the business process of the organization.

So we're applying it, you know, to reinvent how we design chips, how we develop software, uh, how we engage customers, you know, through every business function of the company. And we're starting to see that in, in a big way, happen really across every vertical industry, from retail to telecommunications, to healthcare. Um, and so I, I think it's moving faster than you might think.

I think, uh, you know, en uh, enterprise in some regard has always been a slow beast. Um, it's always moved. The transitions have happened, you know, more slowly than than we would like.

Um, but in my conversations with CIOs, I think what they realize now is it's time to make that shift. Instead of reinvesting CapEx in standard data center infrastructure, take the leap to accelerated computing and, you know, and focus on building agents that address your core business. And, uh, people are seeing, you know, huge, uh, productivity gains and huge improvements to margins that way.

Excellent. Guys, I got one more question and it's for Rahul and Shlomi. I want each of you to answer this question looking into this camera.

Rahul, you're gonna go first for all my friends in ITSM and I'm very good friends with the folks at Idol, right? My friends at People Cert and, uh, Demetrius and they're out in Greece watching this, but talk to all of the ITSM people out there who were worried, is this gonna take my job? Am I gonna have a career?

I just got into this profession five years ago. What is AI gonna do to my job? So I think my thinking is the train has left the station.

If you get on the train, you will have a job. If you don't get on the train and start kind of on the side kind of, okay, naysay, I think you may actually lose your job. The reason is when I've seen the successes, people who have embraced, they are having AI do the job they did not want to do in the first place.

Like summarization after closing every incident, really going after knowledge base updates. Who wants to do it? I've seen people have started going that direction, then they realized, oh, I have now submitted that knowledge.

Why can't I have agent tech do it for you? So slowly they are getting on the train and the train has gone to the next station. People who are left behind, is it not good enough and all that?

Sorry, that is not gonna work. So let me talk to the software developers out there. First of all, whatever Raul said, amen.

Uh, no, no, no. Nothing to add. Software developers, if you remember the days of CICD that you doubted building software with some tools and automation, if you remember the days that, uh, you said that developers in order to be faster, they need to be a bit dirty and not secure.

Those who didn't, uh, um, jumped on the train left behind, and they're not software developers anymore. You have more responsibility and the next generation trusts you to build it, right? Because we are changing everyone's world.

Absolutely. I'll end it with this. I said it before, I'll say it again.

You're not gonna lose your job to ai. You're gonna lose your job to someone who uses AI better than you. Right?

And amen to that. We've gotta learn. It's a tool.

It doesn't replace the spark, the spark that's in all of our brains, right? That cre the creator. But it's a golden age for creators.

Absolutely. And we're lucky to be here. Rahul, Justin Shlomi, thank you.

Thank you for watching. We've got, we've got two more, oh, another day after this. Another half a day here of uh, uh, JFR Swamp Up coverage.

So check it out. We're on Techstrong tv. We'll be right back.

A MD opens up for open ai, Qualcomm disarms arm data centers in space. Qualcomm picks up a new hobby robot, keep away is Veeam buying security. And we're gonna look at the latest threat to Secure enclaves in this week's episode of the Tech Field Day rundown.

Hello everyone and welcome to the Tech Field Day rundown. It is October the eighth, and we are back once again with some more exciting news from the world of technology and some other stories too, because we just couldn't lay off of it. And we hope that you're enjoying National Fluffer Nutter Day.

If you're listening to us in New England, you probably are and are licking your lips. And if you are not, you're wondering what that is. Be very careful when you Google folks.

Uh, but someone who does not have to be very careful when he's doing all of that is my cohost, Mr. Alistair Cook. Al, it's good to see you again.

Nice to be here with you, Tom. And I am always careful what I Google, particularly in a public place. Uh, one of the things I did Google today was to make sure that I was correct about pierogi National Pierogi Day today.

Um, my favorite of the Polish dumplings. And, uh, it's a long way from Poland here, but we have a Polish community who have made me great dumplings here before. Well, the good news is, is that, uh, you should button down your pierogis and be ready for news from all across the globe because we have a lot of great stuff coming your way, uh, not just on the globe, but around it.

However, we are gonna start with probably the biggest news story that happened this week. And that comes courtesy of a MD and OpenAI talking about all things ai because OpenAI announced that they will be deploying six gigawatts of a MD GPUs to power their future AI infrastructure. That deployment is going to start with one gigawatt of deployed capacity in the second half of 2026 and play out over the next few years.

A MD was also quick to offer a warrant for up to 160 million shares of stock to open AI that could lead to them having control of up to 10% of a MD if all of those targets are hit for full vestment and they make the parlay with that weird, uh, punter kicker combo. But the question that I have for you, Al, is six gigawatts of GPU capacity. Are we really measuring that in gigawatts now?

Is that gonna be enough to move the needle knowing that we have seen massive investments from companies like Nvidia? Yeah. And even directly with Open ai, NVIDIA's deal with them is larger at 10 gigawatts than this a and d deal.

Although if you look at it relative to market share of GPUs, this is a pretty significant deal. Uh, I don't think we would expect, uh, a MD to have 60% of the market share that, uh, that Nvidia has in GPUs. And so that comparison is an interesting one to see this new growth.

I think, you know, there's a much wider context here about all of these announcements, and it does feel like we cover this every single episode of the rundown. Somebody is announcing, building vast amounts of data center or vast amounts of power generation or acquiring vast amounts of something. It just, I, no, I'm not sure that it's news anymore.

Only these numbers are huge. Uh, I do wonder whether there's some overlap of the numbers that we're seeing announced that announcing $10 million of this overlaps with, uh, $5 billion of this and, uh, $10 trillion we're getting to was trillions of dollars worth of spent. Yeah.

Trillions of dollars worth of AI spent. There must be a bunch of overlapping in here. And there's possibly some interesting accounting just to make sure that there's new announcements.

Uh, we see this as part of the core. We've commitment with, uh, open AI that we saw, I think we covered that last week or the week before on the, the rundown. Uh, that's a $22 billion, uh, worth of commitment that sit within the, the wider contest of the, the Big Stargate Project to bring, uh, hundreds of gigawatts of data center power for AI over time.

Uh, all of these numbers are outstandingly vast, and I really do want to see vast returns on those investments because there is a huge economic and, uh, even environmental impact to all of these build outs would better see some benefit to humanity from these, these build outs. Uh, the other perspective is that these big numbers typically are spread over multiple years. These are not things that are happening this year.

As Tom said, the the first turn on for this one is, uh, a mere one gigawatt worth of, uh, GPUs. That'll happen sometime in the second half of next year. Uh, for context, there's a ridiculously large amount of power being used by these GPUs and these racks.

And so we are seeing, um, 200 kilowatt racks as being the sort of power density. And so by the time we go to 200 kilowatt rack, uh, one gigawatt is five racks. Um, that's that megawatts No, it's, it's 5,000.

Sorry. That was, uh, I, I missed the, uh, megawatts in between five racks is a megawatts. So to get to a gigawatt, it's 5,000 racks.

Yeah. Okay. That's a pretty big, big environment.

It's, uh, multiple data center rooms. Conclusion on this, there's a lot of money being spent on AI continues to be being spent on ai, people who are betting on this belief we're still very early in generative ai. And that the technology path we're following will continue to require vast numbers of GPUs, vast amounts of compute.

Uh, personally I hope we see some innovation that reduces that demand over time. We've cut up the ins and outs of the legal case between Arm and Qualcomm for a while. Uh, this has been a pretty slow moving law, uh, case of law driven by Qualcomm buying Novia.

And rather than continuing with their own licensing for the, um, CPUs, they took over the Novia licensing, a move that arm says is actually invalid and that, uh, arm wants Qualcomm to renegotiate their original deal. Judge in Delaware has said Qualcomm did not breach the terms and that they, therefore the license through Nivea was valid and owned. Now by Qualcomm, um, of course says we're going to, uh, appeal this and that the, uh, the contract with NI ceased to be valid as soon as they're acquired by Qualcomm.

Uh, does this sound like a benefit for consumers anywhere? Or is this just your usual legal wrangling over contracts between two vendors? I think part of it is legal wrangling.

I mean, there's only so much that you can wrangle on these things. I think what is lining up though is that the judge is really taking a, a microscope to this, uh, you know, licensing agreement and saying, why do you think that this violated the terms when this acquisition happened? Because that's not what I'm seeing here.

And and it comes back to that idea that we've seen a lot in the industry of, if you buy a piece of equipment or buy a piece of software, is there an implied warranty or a user agreement that goes along with that? The manufacturers will tell you no, because they want to get double. They want to get what the original company paid for it, and then they wanna get what you paid for it.

So like, if you've ever tried to buy like a secondhand device on the internet or something like that, and then wanted to update it to the latest software and been told, oh, no, no, no, you need a new support agreement with us. 'cause whatever support agreement they had is no longer valid. You'll kind of know what this this means.

And of course, if you read the agreements that those devices come with, they're very specific about everything in there, how these are not transferable, how this stuff doesn't work. And some of those clauses have been ruled invalid by courts. And what we're seeing here is that a judge in Delaware took a real close look at this particular thing and said, well, why wouldn't the rule that Nuvia had with ARM not apply to the company that bought them?

And, and Qualcomm tried to argue in court. Well, yeah, that's exactly what we're thinking is that when we bought them, whatever their terms were, became part of our terms. And Arm is like, but no, that's a completely different thing and they're gonna have to renegotiate with us.

And I'm sure there was gonna be an additional dollar amount involved in there, because that's what this really comes down to. This has nothing to do with whether or not Qualcomm is allowed to incorporate N'S data center technology into admittedly a lot of things that are gonna be sent out to the customers out there. Uh, they're the new Qualcomm arm processors that are going into these Windows IPCs.

That's where a lot of that new via technology landed. The problem is, is that Arm wants to get paid by Qualcomm for all of that new stuff that they're developing. And if you read between the lines, which is one of the other things that the story is so fascinating about, you'll recall that one of the companies that came out against Nvidia buying arm was Qualcomm because they didn't want this to kind of line up to be like a massive, um, situation where Nvidia would've basically kind of dictated how Qualcomm's business would've run.

I think Nvidia did okay for itself after missing out on that, but I think Arm is still a little mad at them. And so this is like the whole, well, I can't legally cancel your license, but boy, I can make it miserable on you for weeks and months and years and make you jump through every little hoop. And the judge who already smacked the company down in December of last year is basically saying, guess what guys?

Really stop it. So I I, I feel like there's gonna be some rounds of appeals. Some people are gonna take a look at this, there'll be some more back and forth, but ultimately, I think Qualcomm's probably gonna come outta this smelling like a rose, which is really honestly the best thing for consumers.

Bookstore elects Luther lookalike Jeff Bezos has suggested that he wants to put data centers in the one place uncorrupted by capitalism space. Thank you. Tim Carty Bezos says that within the next 20 years, he expects to build gigawatt data centers in orbit that will be operated by solar energy somehow though they can't operate them on planet Earth with solar energy.

Now, I'm gonna toss this over to the real scientists in the room who have pointed out a couple of important things. One, the massive amount of heat that's generated by your average data center is a lot harder to radiate in space, even if you have all of the copper left in the world. And all of those components that are gonna be in those orbital data centers are gonna have to be shielded twice as much, because it turns out that those cosmic rays that cause bit flips randomly on earth when you don't have an ozone layer to block those, they're three times as bad or more depending on which math you look at.

I'm sure that Bezos remains undeterred by all of US news since this project will finally give him a new base of operations against Larry Ellison's Volcano Island Fortress. But Alistair, I asked my question to you is, are you ready to get a contract to launch yourself into orbit to build a data center? And what happens if you forget the rack studs back On earth?

Well, I think there's a a, there's a really important aspect to this, which is to, to understand that, um, Jeff Bezos is sending things into space all the time. And if it's not rock stars, it might as well be data centers. Uh, the challenge, of course, is that these data centers should stay in space for more than five minutes, and that's a little bit more than he has yet achieved with, uh, with his space ambitions.

Stepping back a little bit, uh, motivation here is that when you look at solar radiation, about 90% of the energy of solar radiation is observed by absorbed by the earth atmosphere. That means that in terms of the solar array required, uh, if you put the solar array in space, it only needs to be a 10th of the size of the solar array on Earth. That's kind of some of the maybe logic that's not really working here, because as you say, Tom, that's solar radiation isn't just the good kind, it's also the not so good kind that damages circuitry.

Uh, some of the HPE experiments have seen that in relatively short periods of time and space under two years, um, over half of the SSDs that, uh, HPE shipped up into space had problems. And, uh, that's really gonna be a little bit of a challenge because by the time you add enough redundancy to cope with that, you're adding more mass. And have you really saved energy if you have to use heavier lift, heavier lift than even the, um, SpaceX, uh, Mars, uh, heavy lift, if you have to use as yet unseen heavy lift to get these things into space, are you actually using less energy than if you simply left them on the earth?

Yeah. Jeff, uh, uncle Jeff, as we like to call Jeff Bezos, uh, is saying that this is at least 10 years away and probably less than 20. Uh, yeah, predictions about the future are particularly tough.

I don't see changes in the basic physics here of heavy lift versus, uh, the, the energy that you're saving by heavy lifting being beneficial. Uh, maybe there's something around being in low earth orbit and, and leveraging technology like starlink to get this as being the biggest edge compute deployment. But if it happens, I will be absolutely stunned.

Although, unlike Peter Beck, I will not commit to eating my hat if this heavy lift happens. Peter Beck, of course, is Rocket Labs, my fellow countryman, uh, spaceman. It's time to break out your breadboards.

Um, actually, I've got a big pile of breadboards just over there as well as some Arduinos, because Arduino is gonna be part of Qualcomm, or at least is gonna be owned by Qualcomm. They promise that, uh, Arduino will still be operated independently. The acquisition, of course, was announced in the home of Arduino to run in Italy, along with a new board.

The Uno Q The Uno Q is maybe a competitor to a raspberry pie that runs a real Deion, not a fork. Qualcomm says The move is exciting way to get into the developer community to walk meant its business to business focused. Uh, chip set development.

Uh, the Uno Q uses the new dragon wing QROB 2210 ship, uh, along with a microcontroller on the back. So that combination of high compute power, probably with some ai, uh, chops in that dragon wing, along with a lot of, with a realtime interactions through that, um, microcontroller looks to be an ARM-based microcontroller. Uh, interesting to see that combination together on a board, and interesting to see that as both a standalone computer and, and a development machine.

Tommy, are you gonna trade on your laptop for an UNO queue? Uh, probably not yet. Uh, mostly because it isn't come with a monitor, but I think it's interesting that Qualcomm basically kind of jumped right in here with both feet and admitted.

Why they needed to pick this up is because nobody buys Qualcomm equipment to develop on, unless you're doing stuff directly for Qualcomm. And we've seen this for years, right? Like, I still have a pile of raspberry pie sitting in the corner, uh, doing very random things like I think one of them is Amy, like, shouldn't machine, and, and there's some other stuff, but it was really cheap to get into developing for arm microprocessors by doing that, right?

What was it? You know, um, 30, $40 us. And that meant that a lot of people were writing software to run on raspberry pies.

But as you mentioned, raspberry Os is not Debian. It's a custom version of Debian because it's optimized to run on pies. And so Qualcomm was like, well, on the one hand, we've got people who really should be developing for our platform.

And on the other hand, we've got this really successful group of small device manufacturers who are trying to effectively create this, uh, marketing arm, if you will. So how can we do that? And when you're Qualcomm, the answer is, I don't know what's for sale.

And so they did, they picked up Arduino, and they basically said, you know, we're gonna keep it independent. We're not gonna stick our fingers in it too deep. Um, I, and we're gonna come out with new hardware, right?

Um, the, the smart thing to do probably would've been to partner with Arduino first to see if people were actually gonna buy the Uno QI don't know if they will or not, but I mean, ultimately this can't hurt Qualcomm, because if you can start putting Qualcomm, uh, dragon wings, snapdragon, snap wing, dark wing duck stuff into the community, then what you're gonna get is people who stop writing code that runs specifically on pies and starts running it on the, you know, queue. And by the way, shout out to them for not making this thing a hundred bucks. Um, the, the price points on these things are actually reasonable.

Uh, you can get like the, I think it was the, the 16 gig version is available today, the 32 gig version versus eight and 16, I forget which it's, but there's, there's a small winning by today, and there's a little one that you can buy soon. Um, that is a great way to get people to start writing. And more importantly, the microcontroller aspect of it is directly aimed at Raspberry Pie people, because having that, that, that the way that the microcontroller works on some of those boxes is basically saying you can swap these things in and start developing for us.

And then later when you want to graduate using big boy stuff, well, it's still just Qualcomm chip sets into the hood. So you can already run that. And are you comfortable developing on, on, you know, Debian then?

Great. We, we run Debian, not some customized version. And, you know, I, I think it's a win ultimately for Qualcomm provided, you know, there's a little asterisk there, provided they really do keep their hands off of Arduino, because one of the things that we've learned over the years about the hobbyist and development community is if a big corporate entity jumps in and tries to basically impose their will, we are going to see people defecting away from Arduino as fast as possible.

And they've already had their own challenges over the last few years. It was things as simple as who owns the trademark for the name. So the, the, the community's already kind of a little bit wary of what could happen here, but I'm hoping for the best at this point.

All right, keep your distance chewy. Just don't look like you're trying to keep your distance. And that is according to Rodney Brooks.

And you may know the name because he's the founder of iRobot and an MIT professor Emeritus. He says that people should stay three meters away from any of those modern bipedal robots that you see out walking around. You're probably asking yourself, well, Tom, why should regular people stay 354 barley corns away from these walking terminators?

Well, that's because of physics. And our friend Mr. Brooks says that modern robots are expending so much energy trying to stay upright and stable, that any catastrophic failure of those balanced systems are going to force the energy to be expelled somewhere.

Remember folks, physics is a harsh mistress, so that means that energy is probably gonna go somewhere into the nearest pedestrian. Brooks goes on to say that current robots cannot develop enough dexterity through vision only AI systems. They're gonna need access to other data such as touch.

Now, I will say though, that in the article, there was no mention of the minimum safe distance that you need to stay from away from your robot vacuum. But al do you think that it's probably a good idea to stay away from anything looking for Sarah Connor on the street? Well, depends what your name is.

Uh, if your name is Sarah Connor, then definitely stay away from, uh, whichever kind of, uh, walking full-size humanoid things are out there, particularly battery powered mechanical things that maybe can't feel, both can't feel physically, but also can't feel emotional. There's some interesting challenges in here. Some, some, some thoughts around this.

One of the elements is that we kind of like our robots to be actually androids. Things that look like humans have human form. Uh, yet there are more efficient ways to move around.

I was reading about, uh, putting robots into hospitals. Well, hospitals are designed to have things that roll. So there's no need for your, uh, service robot in a hospital to have legs, and therefore you can get a lot more stability because there never needs to be the, the ability to manage a single point of contact with the ground.

And I think this is kind of where Rodney Brook's argument comes in from and why he's not afraid of his robot vacuum, because it's not standing six foot tall, one legs with maybe 200 pounds, maybe 400 pounds of total mass. I haven't looked at what these things weigh now, but I imagine they weigh a little more than I do. Uh, his concern is that when things go wrong, they go wrong in a catastrophic way.

And that kind of was born out by his early experience of being a little too close to a humanoid robot when it did experience a failure and went down. Uh, he's still alive, so it wasn't too catastrophic. But I'm not sure I'd wanna take those risks too.

Uh, for context, this is, uh, three meters that's 10 of your imperial feet, or just over three yards. Or as you say, what was it, 480 barleycorns. Um, stay far enough away that when it falls down, the bits of fragment that go flying outwards don't hit you.

Or if its arm goes up in the air as it's going down, you get, don't get slapped by. It was the suggestion. And really, you come back to why do these, uh, things need to have two legs?

Why can they not move in a a different form? And why do the two legs need to be placed in the same way that a humans are? Uh, these robots would probably be more stable if they had three legs, but just how far across the uncanny valley would we be if these things with human voices and arms warped around on three legs, or even possibly four, so that there was always three in contact with the ground for stability, all of these things would make these, uh, robots easier to design, easier to operate and consume less power.

But they don't look quite as human. I'm not sure if I need them to look human. Maybe I'm happier if they're safer to be around.

And maybe the, uh, kind of blocky things that we've seen in the past might be a better move than some of the things we're seeing from companies like, um, uh, like Elon Musk, uh, plans to have these blank faced humanoid robots serving us our drinks. Hmm, I'll give them some space. Veeam Veeam's ready to make a big splash in the data security posture Management.

8 billion for security with an eye well with two eyes. In fact, uh, this would help them compete with Rubrik and Cohesity more directly. Hmm, backup companies getting into security data protection.

Uh, DSPM is a big driver for security teams right now, particularly looking to manage the security of data that's being used by AI applications. Uh, Veeam, of course, uh, having moved into private equity has been making a bunch of acquisitions, uh, particularly, uh, picking up the ransomware negotiations group. Core ware, sorry, cove wear, uh, not call.

We, uh, ransomware negotiation group, core ware. I've got that wrong yet again, I'm tripping over all sorts of things. This could be another in a recent spate of acquisitions from Veeam with most recently picked up ransomware negotiation group ware.

Uh, Tom, is this shifting Veeam from, well, my kind of data center infrastructure world into your security world, is this significant for Veeam? It is. And we heard from Veeam at Security Field, a 13 earlier this year.

Uh, Rick Vanover, uh, and Emily did a great job of kind of presenting about ware. But if you look at where companies are now, it's not about those things. It's about creating a holistic view of security and, and what you said, uh, data protection.

And one of the pieces that Cohesity has had for a while at Rubrik has working on that Veeam, is in need of development, is data security, posture management. And essentially what that is, is going through your organization and finding out where all your data is and figuring out how to classify it so you can assign risk levels to it. But there's a lot of other benefits.

And one of those is that if you properly classify data, and I'm not talking about like government classification, where it's like, you know, secret, top secret, and I could tell you, but I have to kill you. Um, this is about, you know, how sensitive is this? What is a risk factor associated with it that's super valuable for ai, right?

Because when you look at the way that we're using AI to do these kinds of things right now, there might be some stuff that you really don't want to have included in your algorithms. And being able to classify that and provide a posture to it and say, okay, anything that's classified as, you know, label X never included in the algorithm, is gonna be super valuable. And companies are gonna be looking for, I'm sorry, customers are gonna be looking for companies that touch that data.

Are you gonna go download a tool that does that, or are you gonna rely on the company that already does look at all of your data, which would be your disaster recovery data protection company? I know which one I would rather give permissions to do that. And so this is Veeam basically looking through the whole market and saying, this is an area that has a lot of synergy with what we're doing, which means we need to be in it.

And in order for us to do that, we need to acquire a company that's already kind of a leader in that space. And this goes back to the comment that you made that, you know, about five years ago, Veeam got bought out by private equity. And I, I don't think I'm preaching to the choir here when I say that the way that private equity solves problems is with a checkbook.

They're not gonna look to develop this technology in-house, or if they did, eventually they'll hit a point where the return isn't worth what they're spending on it. So the next thing they're gonna do is they're gonna go out and buy somebody. Now, again, this is not a done deal.

This isn't even an announced deal. This is a rumor that this could potentially happen. But one of the things that I love about these kinds of rumors is that sometimes they are trial balloons, because if the entire industry starts nodding their head and looking around and going, yeah, yeah, that would be a good pickup for a company like Veeam, that's basically Veeam getting market validation and in some ways, causing valuations to rise.

I'd say stock price to go up. But we know that's not the case. But you know, if, if, if, if the industry as a whole says this is a good move, I think that's something that they're gonna wanna go with.

It's now time for a closer look. Security researchers have figured out how to break enclaves. Hmm, sort of in theory in papers published by two independent groups, um, using the latest exploits.

Now, security enclaves is a way of keeping secret data secret inside your application secret, even from the underlying platform and encrypting data at rest. Uh, both of these attacks called battering RAM and wiretap work by installing an interposer between the RAM and the actual system and capturing the data and replaying it later on. Uh, you need a specific hardware device, you need a specific, um, application, uh, usually a virtual machine to do that replay at this, the particular location.

Tom, does this mean that trusted enclaves where we put our most secure, secure data aren't secure anymore? It depends on how you look at it. So here's the thing, like this was a, a big story, right?

Oh man. Secure enclaves can totally be hacked. And then you dig into it and you realize you had a long way to go from the TPM chip in my iPhone being safe enough for me to unlock it with my face to like levels of like, you know, uh, SHA one or, or MD five hashes being broken.

So first of all, you have to understand that both of the secure enclave, uh, compute mechanisms, intel calls, there's SGX and A MD has one. They call SEVS NM, PSNP, not SNMP, that's a different thing. They both use something called deterministic encryption to store data in ram.

What does that mean? That means that if I encrypt data on one side and transport it to the other side, the cipher text is always the same, right? So like, you know, think, think of something as simple as a ROT 13 cipher.

Whenever I encrypt it over here, whatever shows up over here with the same key is gonna look the same. But you're gonna say to yourself, that's exactly how data is supposed to work, right? When you encrypt it, if you jumble it in encryption in the transport, it's not gonna come out looking the same.

The reason why they did this is for performance reasons. And the researcher who did some of the initial work, I believe it was on, uh, battering Ram said this. He said that originally Intel allowed SGX to run on consumer processors and enterprise, uh, processors.

And then about four or five years ago, they said, we're not gonna let it run on consumer processors anymore because they were, uh, RAM limited, uh, to around 250, 60 gigs of ram. Whereas on an enterprise, you know, Zon thing, we can throw a RAM at it until it falls to the, the core of the earth. But the catch is, is that in order to scale performance to that level, they had to use deterministic encryption.

You can use non-deterministic encryption, but it does include incur performance penalty. And now IT act being able to use non-deterministic encryption would invalidate this attack completely because you can't replay non-deterministic encryption because the, the cipher text is different every time you use it. That's the mechanical thing.

Let's talk about the physicals part. So in order for this to work, you have to have physical access to the hardware device, and it's, you have to install something in the system called an interposer that literally sits between the RAM and the CPU and it intercepts all of the, the, uh, processing that is sent to ram. Then you have to watch a specific memory address for ciphertext, and then you can decrypt and replay problem.

You have to have physical access to the device, right? And I'll recall one of my favorite Microsoft quotes of all time, where if you pop a Windows xp or if you pop a Windows, two Windows 2000 server CD into a Windows 2003 server and boot it, you get unfettered administrator access to the recovery console. And Microsoft's response when this bug was filed was, if you don't have physical access to the box, you don't own the box.

They do. So I could see possibly that a supply chain attack could, could cause an INTERPOSER to get put in there, but that's an awful risk. Folks like you, you've gotta be targeted in order for this to happen.

So you've got, you need to have an interposer. It relies on the fact that SGX is doing this for performance issues, and really all they need to do is turn on non-deterministic encryption, and this whole thing goes away. And battering RAM works on a MD and Intel wiretap only works on Intel.

It is very specific to using Intel. SGX. I don't wanna say there's nothing going on here because obviously this is, you know, this is like spectrum meltdown, right?

At first we're like, oh yeah, pipeline branch execution prediction, that's a problem. Um, and it ended up being something that caused a lot of problems. But the last I checked, I didn't recall anybody using K Meltdown to steal my credit card data.

I mean, Al, do you think that having physical access to the box is a bigger problem than putting an interposer in there to possibly encrypt intercept ciphertext and an encrypted enclave? Well, if, if these encrypted enclaves were on people's laptops, I would absolutely be concerned with laptops and desktops. I'd be concerned about this because physical access is much easier to achieve with a relatively portable device, physical access into a cloud provider's data center, or an enterprise data center that should have good physical security because without physical security, you have no security.

Um, that's less of a concern for me, uh, in terms of the viability of a supply chain attack to put an interposer into every single motherboard, because this would happen either at the motherboard level or potentially inside the actual dim that is, uh, being, being installed in these servers. Yeah, that's not a credible thing, but if there's enough to be gained, a nation state actor might achieve this, would never hear if they did, because anybody who knew about it and leaked the information would be, um, unloved. Uh, and so, so in terms of the, the likelihood of this occurring, my credit card details being stolen this way, I don't see that as particularly likely.

Um, the kind of crown jewels that would be attacked this way would be designed for nuclear deterrent submarines or those kinds of, uh, central banking records that are stored in on these secure enclaves. But these secure enclaves are just another layer of the security onion. They do not stand alone.

They stand in the start with physical security. Start with single tenancy, because part of the replay is that you need to be replaying at the same memory address in order to be able to decrypt that data. Uh, so there's, there's a collection of circumstances that good security practices will allow you to mitigate any risk that is here.

Um, and that's always the reality of, of a security situation is that, uh, you look at what the risks are, you mitigate the credible risks and the cost-effective risks to mitigate that are less credible. Uh, this is probably a, a risk that is fairly well protected by the majority of organizations that are using trusted enclaves. This isn't to say people that are bad at it are not using trusted enclaves, it's just that, well, if you're bad at it in general, you can be bad at it, whether it's trusted enclave and pay a lot more for it.

Uh, you know, that's, that's the reality of the world we we live in, is that any large valuable target is going to be, uh, attacked in in multiple ways. And this is just another potential way. We are a little way away from it being a real danger to causing real problems in, uh, in organizations right now.

I couldn't have said it better myself. Uh, you know, your bank account information is not very important, but a hundred people bank account information is pretty critical. Or the encrypted wire transfer data that you could then substitute your own account information in there, that's valuable.

So I'm, I'm gonna wait with cautious bated breath to see exactly how this plays out. But for now, it's a neat proof of concept and apparently you can build a wiretap, uh, system for less than 50 bucks. So who knows?

But I can tell you something that's way less than 50 bucks. It's free. And that's the upcoming field day events that we have going on.

I have one going on tomorrow, starting at 7:00 AM Pacific. We're gonna be hearing from our friends over at Microsoft. We're gonna be talking about Microsoft security and specifically Microsoft Sentinel.

You probably saw last week they had some big news come out around Microsoft Sentinel, how it's evolving, some of the cool stuff they're adding into it, like, uh, you know, MCP server and a data lake. Well, if those are things that are interesting to you, make sure you tune in. Uh, like I said, Thursday, October the ninth, that's tomorrow.

And, uh, we're gonna have some great conversations with some of their executives like Scott Woodgate. We're also gonna be getting some demos and we're even gonna hold a round table. So that should be real exciting.

And then next week we're gonna be out in Vegas, or at least FST is because he's gonna be at NetApp Insight, which I believe is one of three or four shows that are going on in Vegas at the same time. It's like they're trying to pack him in before Thanksgiving, or some big thing that's happening in Vegas after Thanksgiving. Whatever it is.

Make sure you tune in on October the 15th, because Steven is gonna have some great presentations. It's gonna be a really exciting time. I'm sure he will have a joke or two about Vegas.

You never know. Uh, but after that, Al, you're back with more great stuff. Not in Vegas.

Not in Vegas. I will be in San Francisco in, uh, Gloria San Francisco for Cloud Field Day, October 22nd and 23rd. We have a great lineup of presenting companies.

We're returning to Oxide computing, and we'll hear from Pure Storage and, uh, a couple of other great companies and a, a great collection of delegates as well, joining me out. Um, so tune in for that October 22nd and 23rd LinkedIn or on, on the Tick Field Day website. Also text on TV the following week, we hand back over to Mr.

Foskett. Steven will be also back out in California for AI Infras, uh, AI Field Day, sorry, AI infrastructure is my event. And that's not until next year.

AI Field Day will be in, uh, October 22nd. Uh, AI Field Day will be October 29th and 30th. And Steven has another great collection of companies looking at the things that you can actually do with ai, maybe deliver some value to your organization.

Something else that delivers value to your organization is the Tech Field Day rundown and the news that we bring to you every week. You can catch new episodes of The Rundown every Wednesday on YouTube or in your favorite podcast application. Wherever you find us, please give us a like, and maybe a nice review.

You can also find us on Techstrong tv as well as you can catch many of us, Tom and myself and and Steven across Techstrong and Future on Group programs. We will be back next Wednesday with all of the news that's fit to print, and some of it may involve ai. Uh, until then for myself, for Tom Hollingsworth, and from all of us here at the Tech Field Day Family as wishing you and your family an awesome day.

What, Welcome to another episode of the AI Security Edge, where we explore the intersection of cybersecurity and artificial intelligence with the leaders who are shaping the future of digital defense. I'm your host, Caroline Wong, tech Strong TV podcast features your favorite video series, industry thought leader commentary and analyst research on DevOps, security cloud native and digital transformation. In a podcast format, AI is revolutionizing cybersecurity, both as a weapon for attackers and a shield for defenders.

The AI security edge dives deep into the evolving cyber battlefield where AI driven threats, challenge traditional defenses and cutting edge AI solutions offer new ways to fight back. Our podcast explores real world case studies, expert insights and practical strategies for building cyber resilience in an AI powered world. Whether you're a security leader, practitioner, or AI enthusiast, we hope you'll gain valuable knowledge on the risks, innovations, and ethical considerations shaping the future of digital defense.

Today's guest, lemme see if I can say this right. So there's an American version, which is Francesco Sip, but then I'm gonna try sip sippel. I tried, and then, and then, and then the proper version.

We're gonna try Francesco Chip. Yes. Boom.

Way better now. I'm okay. I'm like extremely proud of myself for that, but you know, just, I think that might have been like a one time thing.

So we're gonna call you Frank. Frank, thank you so much for joining us. Frank.

Frank is a cybersecurity leader, entrepreneur and thought provoker. He is at the forefront of application and cloud security. The most important thing that you need to know about Frank is that he was a practitioner, and now he's a CEO.

He is the founder and CEO of AppSec Phoenix, also known as Security Phoenix, a company that is pioneering contextual, risk-based vulnerability management from code to cloud. Frank has done all sorts of cool stuff at HSVC, at AWS, at the UK and Ireland chapter for Cloud Security Alliance. He is a professor at Ions.

He is a multi award-winning podcast host. It's actually weird for Frank to not be the host right now. He's a regular keynote speaker, he's an author.

He writes books, white papers, articles, and, uh, he's also a self-taught artist and a former professional skydiver. So if this is the first time you're meeting, Frank, I'm so excited for you because you know what, Chad, GPT, uh, there, which is actually like, that's an AI use case, right? Um, if this is the first time you're meeting Frank, or are you in for some good stuff because there's so much good stuff, Frank, welcome Caroline, as always, you shine.

Thank you for having me. So Frank, What did you find? All this stuff, Everyone, uh, literally it's chat, GPT.

So everyone on this podcast, I, I like to ask the same questions, but, but you're not like a, you're not like a typical podcast guest. Not really. And so I'm gonna ask you a different question, which is tell me what you actually really think about all this AI stuff.

Tell me the real brutal raw truth. It's a bubble. Uh oh.

Uh, but is a cool Bubble. Okay? Okay.

Tell us more about this bubble. com or not as experience. Instead, right now, we have organizations still trying to figure out how to prioritize vulnerability, how to do cloud, how to do software, while attacker extremely enthusiast about, Hey, let's use this technology, or let's weaponize the model that are trying not to do it.

And I think Tropic has published, uh, a recent playbook on how attacker are creating new method and way, and they, of course, they're trying to stop, they're trying to ban them, they go through. But we start seeing case where LLM are weaponizing vulnerability or are being used to attack ransomware. So AI has lowered the barrier of, of, of access for cybersecurity professional, but also for attacker.

And we were overwhelmed before, like I think right now the difference between the dotcom bubble and right now is we're seeing this technology put exciting, but we are seeing it as faster growing as a weapon and as any new technology, we are seeing the rush to market. Of course, Trump insecurity read us as MCP because API security wasn't hard enough, so we needed to have GraphQL. And one of my good friends is saying, I love any GraphQL because I can hack the way through it very easily.

And because that wasn't sufficient, we had to create MCP, actually, we released our MCP server and we shut it down for security concern. I'm proud to say it because we did a threat model on that and saying, that's not good enough. But how many people out there are throwing the MCP out in the ward and saying, yeah, it's secure enough, right?

Frank, I I have to pause you for a moment because there are folks listening and watching who know what GraphQL and MCP are good for you, and there's people who don't. So for the folks who don't give us a little bit of background, talk to me as though I'm my 75-year-old mother-in-law, or my 10-year-old daughter. I think you might be right.

I, I get over excited about technology sometime and I think that everybody lives in the world of stuff that my brain leaves. Um, sometimes Only the really smart ones, All the crazy one. Uh, but thank you for the compliment.

I think when we look at the internet, we had the history of API that were soap XMLs, a very ancient way to pass data through a system that expose a web interface, and then we kind of settle on rest API. That is the standard method where we taught really, really well and long about how to secure those things, how to create that entity. So I think rest API has been around for very long time, but for rest API, you had to create basically endpoint for everything that you want to do.

And that is means development. So some of the dev team has said, why not throw caution out of the wind and open everything to everyone? Just query whatever I want and I expose anything that I want.

Because that has worked out well for us in the past. So that's was the history of GraphQL that you can secure, but it's really difficult to constraint or provide access control because fundamentally you can tell, gimme the information about this, this, and that. And GraphQL would say, gladly, here you go.

Uh, you have the permission to see that stuff. Hopefully you have your pass through credential or pass through authentication configured. Most of the time you probably don't.

So you create just access to your data lake and if you're lucky, you just see what you wanna see. Um, but it's very difficult to control. Now, MCP have been built in a rush on a protocol that has two or three version, and eight, two A was the evolution of the MCP protocol, but authentication was nowhere to be seen.

And all to authentication token being passed through or authentication and authorization have been kind of left in the world. So we're seeing MCP being exploded up, down left and right because it's a new technology and because it just rely on not very strong foundation of authentication and access control. And that's one of the reason why we shut down ours, because our API will build with Phoenix security with specific method in mind.

So we put, uh, an MCP server in front of it, and it gives, it gives you access in a different way that we want. And we expected, so we did a, like any security folk would do a threat modeling exercise. We deem the things not secure enough and we say, you know what?

Let's leave the hype to the hype. And I'd rather not get hacked than be late for a few weeks. Um, and that's what we did.

But I think we won the few that actually take that hint. That's so interesting. Uh, humans want to use technology to share information and then they end up sharing it with people that they didn't wanna share it with.

And if you're in intentional about putting some cold controls in place, then it takes more time, it takes intentionality. Um, and now we have not only automation, but we have ai. So the problem is just worse, more data, more places for that data to be more places in our supply chain to poison and to steal information from.

And so Frank, I think yeah, please. When You have, we have, I've been thinking about this very hard and very strong, like why LLM seems so attractive. It's like, why is so easy to get caught into the perception that's, we have an answer?

And the answer was there is the fact that LLM always give you an answer despite that it's good or wrong or whatever, or whatever precision you have, you always could get an answer. It might be wrong, but you always get an answer. So it feels that you making progress despite that you are actually making progress or not.

And that's the intoxicating element of LLM. You don't know anything about API security, I'll ask LLM to do, teach me about API security. You don't have context.

You haven't asked us specific things, but it will return you with some stuff. Um, hey, I have this code. What does this code do?

I wanna do these particular things. It will give you an answer. It's probably wrong.

But that's why the excitement, because the barrier of acquisition have been lowered, the fact that it doesn't always speed the right information is a different story. And hence why people that understand how AI was built. I was building bias network and neural network back 10 years ago when AI wasn't cool.

And me and my co-founder understand really well how AI was built. And LLM is just a variation of an ai. And if you understand how it works and how to ask the right question, you become a superpower because it really 10 xs you.

And I think I'm, I'm surprised by the kind of things that if asking the right questions, it will give you the right answer or it will speed up your work, but also can slow you down tremendously. Or it can create a generation, I think of no brain coder and as an industry, I mean you in threat mode con always, we, we, we go long time, me and you, and we are seeing the industry kind of trying to make an effort. I think right now we are creating a generation of people that don't think securely or they don't even understand what a vi vibe coding.

So that's a little bit my fear of creating a generation that doesn't have the understanding or the baseline understanding, but just go with it and vibe with it. And you can vibe secure coding. I mean, our good friend Jim Monica has created a whole training about vibe coding securely.

And I think you can, you just need to know what to do and what to ask and how to ask it. And you still need the principle to be in there because AI will not magically secure your application. So Frank, uh, what I hear you talking about is a comparison.

Uh, there is kind of like the no brain way to use ai, and there is on the flip side, a very powerful way to use ai. And so my question for you is, what advice do you have for our listeners to be, not the former, but the latter? How can we all learn to be the best users of AI and not the no brain ones?

That is a great question. And for that, we've broken our manifesto. What, what, what, okay.

Uh, It's not yet public. Okay. So we call it, okay.

Oh my gosh, ai, ai. How is Everything? Ai, human plus Ai second human first, the manifesto tell us everything.

So I've been thinking a lot about this, and I think with all this hype, we tend to, we tend to place AI first you see a lot of company coming out and saying, we are AI first. AI is gonna solve all of the problem in the world, and it's so cool and it's whatever. No, AI is just a tool.

And like blockchain was just a tool. Let's try not to create solution before we have problem to solve Engineers. And I know, right?

But in general, if you, if you treat AI or LLM or vibe coding as a technology, as a tool, and you learn how to use it, you become really powerful. And I think in few years that's what's gonna distinguish the people that talk about by coding LLM, but they don't know how to use it to people that have experience and know when to use surgically technology for that experience. And hence why we say human first, empower by technology like an LM, like a chatbot, like an AI tool to 10 x their capability.

But ultimately you'll never be able to fire an ai. So decision will never be able to be delegated to an agent. But an agent can 10 x your engineers.

So if you train your engineer well, junior and senior to use technology in the proper way, then you have a force of nature. And I think our attackers have understood that. Well, first, some haven't.

Some vibe codes. Write me the, um, what was it, what's a ransomware letter for the FBI for the director of FBI? Because we have all of that data, uh, and somebody has came up with the same kind of things with Google without any proof and without proofreading or so on.

But in general, you have people that understand this technology and they want to use it and AI second, and you have people that put AI first and they will be left second. Yeah. And hence the manifesto.

You know, I'm so excited for this because it truly is the message the world needs to receive right now. You know, a year ago, and still today, every board on the planet wants everyone to use AI for everything. You know, every engineering team is being told Use ai, use ai, use ai.

No one is talking about how to do it properly, how to do it. Well, boy, is there a difference between doing a thing Yeah. And doing it Well, I, I can't wait.

I can't wait. Who, who, who, uh, who's coming up with this manifesto? Tell us about the creators.

So I generate the first idea. I sent a few of the leaders that you well know, Azar, a few others that have done their first pass on it. Um, few other CSO and, uh, thought leader as well have contributed on it.

We'll have the full, I think we have 25 right now in the us. We try to mix practitioner and CSO alike and non technologists to actually come up with a message that was sustained by both practitioner in security field leader, a CSO in the security field and non practitioner to actually write something. And we wanted to keep it purposely short with 10 commandments that really say, think about these things securely and think about this as a technology.

Like that's the underlying mean. We can go through the manifesto, but that's the underlying message of the manifestos. Like, use tech, use this technology as a technology, use it wisely.

Like by code. Absolutely by code the hell of things. Um, as a CEO, I push for AI adoption, not AI first, but AI adoption to all my engineering community.

But also we have guard rails and we have methods of embedding things. And we are actively researching how to insert secure prompts in the by coding thing. So they will over return a secure by coded message or prompt.

Like that should be the core of what we do and the core message of what we do. It shouldn't be, if you don't use a vibe coding tool by Tuesday, you're fired like some CEO have put. Yeah, I think that's a wrong message because that's, that create that people will adopt, people will adopt and people will make mistake because it will delegate thinking to the technology.

Well, this should be a thinking aid. It shouldn't be an outsourcing. I might be, I'm popular in this opinion, but I rather us going forward with the eyes well open rather than creating, what was it, the movie Terminator.

Sorry, I had to throw it in there. You know, Frank, what I like about this is what I'm not hearing from you is I'm not hearing any fear. What I'm hearing actually is a sense of empowerment.

You recognize the power that we have as humans. You know, do we store tremendous amounts of data in our heads? Yeah, we do actually.

You know, do we have decision making? Do we have discretion? Do we have judgment?

Yeah, we, we do actually, you know, and so I'm delighted to hear this sort of elevation appropriately of yeah, the human, uh, and who is in charge, right? The human or the machine better. You can Fire, you can fire a machine like ultimately comes down to that.

Like you wouldn't be angry at the machine because it does machine job or it doesn't error or it has a bug. Ultimately, technology is technology. And we need to recognize this as a technology.

That's we, that's what we, in Phoenix, we created our AI agent as copilots that aid decision making process, but empower people to make those decisions. Ultimately, we present three remediation plan. We don't know better than the engineer.

We give you guidance, we give you insight, we give you direction. And we say, based on this, and we explain the reasoning as well based on this, this is why we doing specific things. But then if you think that fixing things by a specific asset or fixing things by a specific threats attack vector is better, choose that remediation method.

So we want to empower instead of replace human cool and security engineers, I love it. And a lot of people are scared right now because they, this technology feels like is, is AI is gonna replace or go or come for my jobs? If that's the fear, then you're in the wrong job.

You need to elevate yourself to use technology. I think that's where the fear come from. And you have I think two sides of people that fear a technology because they feel overwhelmed.

And by all mean this, uh, scary technology because it seems to be able to do everything and nothing. So either you embrace it or you be left behind. And that's the hard truth.

So it's better to embrace it, use it securely, and be at the front edge of this. But if you were doing spreadsheet yesterday, I'm sorry, this will be replaced. Yep.

Hard pill to swallow. Uh, but I agree and uh, Frank, as we're kind of beginning to close up our conversation today, for folks, maybe today's the first time they've learned about Phoenix security, tell, tell folks about Phoenix security. So, long story short, we were a bunch of practitioner.

They were leading AppSec Cloud set transformation in most of the banking world. And we wanted to solve a problem that is how do we align executive expectation to engineering action? One of the frustration that we had was when we talk to engineers as security practitioner and as security leader, we tell them, you shall secure your system.

And when they look at us and say, what does that mean? We don't have an answer, or if we have an answer is, well, you need to fix your vulnerability by SLA or you should do threat modeling. Okay, teach me, I dunno, this is a template to use it goodbye.

I don't have time. We don't have scalability. So we wanted to empower, first of all, engineer to understand this is what security expect of you.

And then we wanted to align that message with business expectation. Because if it's not important for your boss as an engineer, you're never gonna be giving attention to a particular problem. So we wanted to solve the problem of security across application security and, uh, cloud security.

That is called vulnerability management. That is a problem that we had for 20 past years, and we wanted to solve it from a business perspective because that's the only way it actually work. And then in that journey, we evolved that with asset inventory.

That is also another big problem that we discover in the journey, saying, if we don't know who needs to fix what, how can we tell them to fix stuff? So we open source our CMDB, yamo based CMDB to empower every engineer to declare this is what I own, and I don't have to log into one engine 1999, uh, black screen with green line system. I can just declare a yamo file, inpo.

And that's automatically configure Phoenix to say, this is the stuff that this team owns. So if they have vulnerability and you expect them to fix it, we're gonna notify exactly who needs to fix what, where, and ex tell them why it is important. And in a nutshell, that's Phoenix.

That sounds really cool. Frank, if you could go back in time and do the job that you were doing at HSBC, what would it have been like for you if Phoenix Security Technology had existed? Well, it's funny that you asked, because that's where Phoenix was born.

Incredible. We created that for ourself in there because we had that frustration because we couldn't translate an executive saying we should do security. An engineer saying, what does that mean?

So we created a way for executive to report this is the percentage of security that we want to decrease. This is the risk level we want to go. This is the amount of money that we wanna reduce in terms of direct and indirect impact.

And that very high level message that a non-technical, um, or risk base executive can express, could be translated to engineers saying, this is the vulnerability you need to fix. This is where you need to fix. And we as security were coming and saying, look, if you look at this library, these system, these things, you actually maximize your risk reduction.

So you will look way better for your boss. So instead of demonizing engineers who were coming and aiding them to get to their target faster, and look, that was four years ago. So it was a very, um, early stage Phoenix.

But that's what the gamification from a business perspective and from an engineer perspective, is what have enabled us to move from resolution time of 290 days to 20, 30 days. Nowadays, it's not sufficient anymore because I think with the latest data that we've seen, expedition time fluctuate between three minutes and seven days, depending on what kind of data source you look. So 30 days is not anymore for critical, but if you don't know who does what, probably you are over a year of remediation.

Yep. Frank, last last thing that I'll invite you to consider doing with me. I want you to teach me how to say your name properly.

Can we, can we try this together? Let's, let's try, please say it and I'll see if I can repeat. So I usually, it's a funny joke and my partner always makes fun of me because I say I go by Frank for friends.

And then if somebody doesn't call you Frank, it's like, does that mean that they know your friend? So I don't realize it's, it is, it is something that is ingrained right now with me. But if you wanna try in the Italian way and you did it beautifully, actually, uh, it's Francesco chip.

Francesco chip. That's great. Yeah.

Okay. I'm so happy. Um, gosh.

Thank you. Thank you So much. You honor Italian now.

Thank you for your time today. Thank you for your wisdom. Thank you for the work that you're doing for our industry.

I cannot wait to read this manifesto and tell the whole world about it. Thank you. Brilliant.

I think you very man needed. But thank you so much for GE having me on this side of the podcast. It's my pleasure.

Folks. Techstrong TV podcast feature your favorite video series, industry thought leadership commentary, analyst research on so many topics including AI and cybersecurity, but also DevOps, cloud Native digital transformation. Uh, come on over to Textron TV podcast to find all of your great content.

This has been the AI Security Edge. I'm your host, Caroline Long. Thanks for being with us today.

Our online interactions include audio, video, and sensor data, but most AI applications are still focused on text. This episode of Utilizing Tech considers how we can integrate multimodal data with ag agentic applications. With our conversation with Vka Gupta, founder and CEO of Aperture Data, Frederick Van Hern and myself, Steven Foskett, welcome to Utilizing Tech, the podcast about emerging technology from Tech Field Day part of the Futurum Group.

This brand new season focuses on the practical applications of ag agentic, ai, and related innovations in artificial intelligence. I'm your host, Steven Foskett, organizer of the Tech Field Day event series, and host of utilizing Tech Now for nine seasons. Joining me this week as my co-host is, uh, Frederick Van Herren, who's been present for a lot of those seasons.

Welcome Frederick to the show. Yeah, thank you. Once again.

I'm, I'm, I'm here as a co-host, so I name is Frederick Van Herren, I'm the founder and CTO of which is A HPC and an AI consulting and services company. And Frederick and I have been talking about practical applications for AI for a long time, but one of the things that always sticks in my craw, I'm not sure what a craw is, but something sticks there, is that when people talk about ai, they too often focus only on text. Basically, it's chatbot or bust.

And that's great. And in fact, there's a lot that you can do with text, but text isn't the whole world. Uh, and Frederick, I mean, your background in ai, you, you, you didn't start with text.

No, definitely not. You know, my, as you know, my background is in speech and, and that's the language we use to communicate with people, but we have to understand that people are more visual than learning from text. So a multimodal approach to ai, it's definitely something we're looking forward in this agentic AI world.

Absolutely. And I think that, um, you know, just regular people who are gonna, uh, be sort of wondering why are we, why are we always talking about, you know, documents and books and webpages and stuff? Why aren't we talking about literally everything we interact with today, which is, um, audio, which is images, which is video documents and so on.

And that is why, uh, we have an exciting guest to kick this season off. Uh, today we have, uh, Vish Gupta, uh, founder and CEO of Aperture Data, who is somebody that I spoke with, uh, earlier this year, and they are really focused on multimodal data. Welcome to the show.

Thank you so much, Steven and Frederick, uh, really happy to be here. So tell us a little bit more about your background and yourself and, uh, and what you're focused on. I'm happy to.

So yeah, as you mentioned, um, right now I'm co-founder and CEO of Aperture Data. Uh, prior to that, I was at Intel Labs for over seven years, which is where we started working on this problem. And, uh, you know, just take looking at it, um, from a researcher standpoint, and now through the journey at Aperture Data, it's been more like product and user standpoint and businesses standpoint.

Um, for my background, um, I have a PhD in computer science from Georgia Tech and a master's from Carnegie Mellon. And I got my undergrad in computer science from Biti in India. So it's been, uh, you know, one part of life where it's a lot of, you know, computer science, deep research, working, really like, you know, uh, on underlying, um, uh, systems hypervisors.

We were one of the first teams to virtualize Nvidia GPUs back when it wasn't even so, uh, popular in, uh, data centers, uh, to, to be, to be offered in cloud environments. Um, and then now this has been a completely different part of the journey, you know, by sometimes, um, miss coding too. Uh, but it's a lot more about people than it used to be before.

And most importantly, it's a lot about very exciting use cases and applications that have emerged in this last decade. Just as you know, we have witnessed the progress of machine learning from, just from the very basic, like, you know, the very first image and that thing that make it like, oh, now we can actually automatically understand what's in an image to now AI agents trying to like, you know, order our plane tickets for us to plan this perfect vacation. Well, it does seem though that, um, the progress of ai, you, you're right, that it started, as Frederick said, it started with, um, with speech processing, uh, for the most part, uh, it, it, a lot of the applications early in the utilizing journey.

Back when we started this podcast, we talked about, uh, processing images and detecting objects and, and, and processing video and sound and all sorts of other data sources too, um, you know, bio, um, mechanic information from sensors and so on. But it seems like the prevalence of large language models has just crushed any kind of discussion of anything that's not text. Are, are you seeing that as well?

Um, I think a lot of the practical applications, you know, the, the approach I end up seeing for people a lot of the times is, well, they're asked to go use ai. So this is like, you know, you kind of have to, you know, we work with a lot of, um, you know, medium to large enterprises, some startups, and we see this very often. There's always this like, how can we use ai if, if they're doing it right, it would be more from the perspective of, there is this business problem, can AI help us here and then watch the process?

But then sometimes it's like, we gotta get on the AI bandwagon. There's a lot of funding for it, right? So there is like a spectrum of people.

Uh, but the common thing is like, okay, look, uh, especially when you think about larger companies, data is very siloed, right? Especially if they're collecting, if they're going beyond simple tabular data, going beyond text, that means sometimes it'll be like images, videos, audio, they get organized in places that most of the company doesn't know, like few people in some team that will have access to it. So just the process of bringing things together and then making sure that the models are up to par, and then can you actually take all of that and combine that into, um, a model that can give you the right answers?

That's a very, um, intensive process and which is what makes it like, okay, well let's prove the value with text first, uh, and then we'll see. Right? Uh, unfortunately though, the problem is in some cases text is sufficient.

A lot of the times people are like, you know, let's say if they are just, uh, uh, there is an example where you have a lot of PDFs, and granted it took some while to start parsing PDFs to the level of, you know, actually understanding tables and images within PDF two, it's not, it looks simple, but it's not always, but like, you know, if you're trying understand a lot reports you do internally and enable a chat bot chat, that's the kind of application that you can pretty easily start off with. Um, uh, and, and, and, you know, so you start, if, if you start seeing the ROI good, but if it's the sort of application where a lot of information was stuck in these other data types and you started just with text, it is quite possible that people arrive at the wrong conclusion that AI doesn't work. And I've seen that through a lot of, uh, you know, even before we got to the whole rag and egen stories that we are like Gentech world that we are in today, even when people were just trying to do like, you know, let's say e-commerce personalized recommendation, people wanted to use visual similarity search to recommend products that look like this.

Because, you know, we're very visual people, like you said, uh, we're, uh, we look at something and that's what attracts us, not the text description of the product, right? Um, and a lot of the times teams would start building it, but they didn't have tools to be able to query and see what their image data sets look like. So they would just train the models, you know, with partial knowledge, not do the best job, and then the recommendations weren't as good as, you know, just based on your friend bought this, and so you should buy this sort of stuff.

Um, and so the conclusion used to be, well, AI is not helping us. Uh, and, and so that's why I kind of warn in terms of like, you know, it's a text is a good start, but there are a lot of cases where you got to bring in the other signals and it looks very daunting because the tooling also has been pretty broken. Um, but that's kind of why we are here.

Right. So you talked a little bit about, uh, multimodal. I mean, it's, it's already difficult enough to build models just for text or audio video, um, let alone the different data types, right?

Video notoriously are much bigger, binary, very difficult to analyze. Uh, text is easy to read, but much smaller than video. So how do you deal with all those different data types and those different models into, into one final model, so to speak?

Um, I think I, I mean, I couldn't speak, so the way I look at multimodal data and what it would take to unlock everything that it offers, I look, uh, at it at, in, in three, three sections, so to speak. One is the models, what you're referring to. And you know, there are a lot of vision language models now that are doing really well.

I mean, I'm recently reading all about like, um, generative on the SOA side, but also in interpreting like the, uh, if you look at some of the Gemini output and stuff like that, they can really do a very good job understanding what's happening in image video. The second aspect is processing. Well, there is no lack of processing today, I might say.

Um, I mean, Nvidia really has changed the name of that game. Uh, and there's a lot of other, uh, inference provider providers and, and, um, some more niche companies, uh, coming up in that space. Uh, and then the third aspect is data management.

And I feel like this is where the biggest gap exists. If, if you look through the evolution of machine learning, you know, we used to see all these papers where, oh, you know, we are now able to detect like this tiniest bit of, uh, like a dog face in the image. And then you went and deployed it in like a medical imaging sort of scenario.

And I literally have an example I ran where the brain lobe, uh, like the brain scan was classified as a telephone lobe. Um, so, you know, so those sort of things, you, they happen because like data has always been the differentiator. The better data you train with, the more representative data you give to your model, the better the outcome is.

Um, and it remains true to this day, but unfortunately, the solutions are still not there yet because changing data systems is a very involved and very complicated process. Um, or it, it doesn't have to be, but that's how we've always seen it. Like, you know, I mean, there's so much, and like people think so much in terms of SQL and relational tables.

Sometimes I run into people, it's like they won't use anything that doesn't support sql, but that's not the right approach. The thing is, what solves your problem if, if AI needs to see all of the data, you need a data foundation that allows you to put all of the data, make it searchable, and make it easy to navigate. That has always been our driving principle behind this, because if you wanna go, um, and you know, earlier we were talking about going from shallow intelligence to deep intelligence.

Um, we are, uh, you know, like I was saying, the models really are very advanced now, the processing power, you know, it's growing constantly and, and, and accomplishing a lot. If we solve the data problem, if we build the right foundation for data layers instead of still cobbling together a bunch of tools that make you inefficient, that, um, create inconsistencies that don't let to you scale as much as you can, you're still never gonna fully go into deep intelligence territory. Yeah, it, it's, it's so true.

And, and what I'm seeing unfortunately is that a lot of ag agentic applications are still focused on structured textual data. In other words, if we're gonna have this process passed data onto this next process, um, in many cases what it's doing is it's devolving, uh, you know, let's say visual data into a, in, in some cases a large set of texts that describes that visual data in either, uh, freeform text or in, you know, structured, um, data and then passes that to the next thing. Um, is it possible for agents to pass the actual image or some abstraction of that image between, uh, agents in these systems?

It, it absolutely is possible, right? So, um, as we were building, so, you know, as you know, the product that, uh, that my company, um, offers is called adb. It's this unique vector graph hybrid database that we have purpose built for multimodal ai.

So there was always this aspect of, you know, if someone says, I wanna find images similar to this image, there's of course the vector search angle, so you need embedding generation and things like that, but how are people gonna give you the image? Because they would have to give you the image to put in the embeddings, and then you can do vector search, right? They would have to, um, and let's say you were going even further, you wanted to find video clips that had, let's say, kids playing in it.

You would have to be able to understand components of the videos, um, you know, generate embeddings from those and be able to search through them. Uh, but it's not just the vector search part, because what ends up happening is, let's say you want clips where, um, uh, kids are playing in the video, right? Um, maybe the entire video is 30 minute long and there is only like a two minute section in which the kids are playing.

A true multimodal AI database should allow you to search, decode the video, and go into those two minute part and just transfer that. Now, why is that important? Because, I mean, imagine videos, like you mentioned earlier, videos are really large.

Are you gonna be transferring the whole like, you know, 30 minute long video between different components that need to operate on it, or do you wanna just take the two minute sections that are relevant and pass those along? So that's like, that there is, there is this whole efficiency angle to it and, you know, not having to wait for hours for something to happen, something happen that can only be enabled when you introduce true multimodal understanding in your database. And that's, that's kind of what we did because, um, you can literally say, I want all the video clips in which a person was smoking or not smoking, uh, and I want them returned to me in thumbnail size.

This is one query to Aperture db. It does the decoding, it tricks out the parts that are interesting and it, you know, you know, bundles up the clips and sends them to the, uh, to the next stage. Um, and you are not duplicating any of this information because remember, you gotta think about scale.

We are gonna operate, we we're operating on petabytes or maybe even zetabytes of data, right? Um, and when we represent videos in our database, that is the original video file, but then we very smartly use the graph structure that we have to represent all the regions of interest in it. It can be interesting frames, it can be interesting clips in it.

It's the same logic with images. You know, sometimes, um, you might, your cameras might be really high resolution and capture a very wide angle, but all you care about is that person that's standing on the street. Why are you transferring all those pixels between the different stages?

Um, so to your, to your original question in terms of why can't you transfer some of these other data? Can you transfer these other data types? I think one is the protocol that allows you to define the stuff, and I think there is still some room to improve.

Like we had to, um, come up with a different query language to support all of this. Uh, we actively chose not to implement into like a SQL or a cipher based query language because they were too restrictive in what we were trying to do. Now we have built plugins to make them compatible because a lot of other tooling lives in that world, but, uh, we started out without hindering ourselves, and, uh, we had to introduce the exchange, like, you know, okay, this is how you are gonna give us blobs of various types.

This is how we are gonna demarcate one blob from the other, and the metadata we return is gonna tell you what the rest of it means and things like that. So, um, we are able to do it. Um, and, you know, we, um, work with PDFs, audio images, videos, and, um, it, it, it works great.

And of course, in the backend, uh, we've introduced the whole performance and scale and, you know, understanding that these data types are different. You have, you know, more parallelism requirements, there is less dependency among, you know, individual parts of the data. So there's all that stuff that goes into the architecture to make it high performance and efficient.

And then there is in the protocol to, to enable like to, to define that language. So it's, it's very much possible to do it right. Yeah, data movement has always been a problem, and as, as, as people collect more data, I think the data movement by itself will get worse and worse.

So how do, how do people interact with your platform? Do you integrate with frameworks or orchestrators, or how do, how do people use and consume the platform? Yeah, I mean, you know, we started out with a database.

No one wants to really think about a database. It needs to be hidden behind stuff. Um, and so yeah, we have, um, uh, you know, originally when we started, because we were looking at a lot of, um, training and inference sort of use cases, we integrated with sense of flow, vertex, ai, these sort of, uh, frameworks.

Um, then we, you know, with the rag in the rag world, basically we introduced Lang Chain LAMA index integrations, and now we are looking, looking into ENT memory, uh, frameworks to integrate with them. Um, we also do, you know, so ADB has grown from just being a database into this entire platform where, um, you cannot just manage and search the data. But, you know, we have introduced workflows to make it easy to upload data to, uh, generate embeddings to extract information.

I mean, we have a workflow that, you know, you give a URL and outcomes are at chat bot. Uh, you don't have to worry about segmentation mo like, you know, embedding models and things like that. Um, so, and, and we have these various, like, you know, MCP server plugins, SQL server plugin.

So that has grown, uh, now into a platform so people can interact with ADB directly on our, uh, cloud platform. They can use Community Edition, we can do a VPC deployment. Um, but we also have our own ui.

And I'm actually really pleased recently with all the developments that have happened into our UI because, uh, you can literally go to, you know, one of the tabs in the UI type a text question, and if you've like, you know, ingested the data and generated embeddings, it'll show you, okay, these are the images that match, these are the PDFs that match, these are the videos that match and all of this on one interface. And there is still so much to, to, um, improve there. Yeah.

So the, the, the, the workflows or the plugins are, are kind of starter kits and I guess for, for, for the, for the consumers. So can you talk a little bit bit how the platform works with, uh, enabling autonomous and semi-autonomous AI agents? Right.

So, um, there are different ways that you can go about it. Like, you know, a lot of the agents behind the scenes when they want to interact with data, they basically might, you know, just do vector search queries and then, you know, implement their own LLM like feed, like do the semantic search and feed things to the LLMs and then generate the responses. Um, so that's like very fundamental way, which, you know, you can just use the vector search support we have.

You can enhance that with graph rag, sort of, you know, like rag improvements to start including the knowledge you've contained in the graph. But where we are seeing this go is essentially introducing this memory interface, uh, because if you look at, you know, the memory frameworks, there are, there's the component that actually takes, um, user log user questions and extracts preferences and, and, you know, relevant personas and stuff. But underneath it ends up storing this in either vector databases or a combination of vector and graph database or just simple text logs.

And ADB is perfect for storing all of that stuff. I mean, the throughput and latency we offer in terms of updates and queries, it's phenomenal. Um, and so it makes a really, you know, good foundation.

And so now the, the thing we are working on now is like, okay, what's that memory layer, right? Like, you know, we start by integrate, we, we will basically integrate with some of the frameworks already out there. Um, so the agents can really like, make use of the memory and scale, um, through what Aperture DB offers.

So I love this talk of, you know, moving beyond text. I mean, that's the, the, the premise here at the beginning. Um, but, um, I wonder if you could help us with some or examples or some ideas about moving beyond video too, because of course, multimodal data, it doesn't just mean video, it means all sorts of data types that are, you know, all varied.

So what other data types beyond audio, video, video and obviously text are, uh, people looking at with agentic applications, and what are some of the use cases for that? Um, well, you know, I mean, uh, Stephen, as we talked about, a lot of people are still on text. We are not even at the audio video stage yet, but, you know, so there is a lot going on with voice.

So there's a lot of audio information. I think people have realized there's a lot they can even, like, you know, even before getting to voice and, and videos, um, there's a lot going on with PDF because they, I mean, you know, if you, um, we, we, we all create so many reports, right? And we like to put tables to summarize, we put charts there, um, we have these pie charts and you know, sometimes we put pictures to show the way, like, you know, how our architecture looks.

So there is a lot that goes in and parsing PDFs in itself is, uh, and, and extracting information to start making sense, and, you know, at what b um, parts do you, uh, how do you segment it and things like that. All of that stuff involves a lot of, um, work. So I've been seeing a, like people who have managed to go beyond text, a lot of the times it's like, uh, actually this is the kind of progression.

You start text vector search, right? Then you start realizing, well, you know, your text is giving you some more relationships about things, and so can we connect and start, you know, utilizing the relations around it? So it naturally kind of progresses into well graph sort of notion, can we build a knowledge graph?

Can we use that information to improve the responses? Then it moves into like PDFs and that auto automatically gets into like parsing images and stuff. Um, of course voice AI companies.

I, I think Frederick, you would know a lot more on this one. They are, they are starting to get, uh, you know, voice in my understanding started with like, let's convert this audio into the transcript, and again, go back to the vector search where I think there is an increased understanding around like, Hey, if you did that, you lose the emotions, you lose the, you know, background information. And that's sometimes really important.

Um, so you go beyond that. If you get two videos, there are some cases, especially in, uh, medical imaging sort of cases, you know, where the scans, um, like, you know, nowadays a lot of the CT scans or ultrasounds can be pretty like, uh, 3D formats that can be, um, the neural scans are in a different format. Uh, so when you go into more, uh, specific, like more domain specific use cases, then the file formats start to be different.

So that is, um, you know, can you understand, um, the medical imaging file formats and start and, and enable the medical copilot sort of use cases, right? Because I mean, patient information is naturally multimodal. Uh, then there is, uh, satellite imaging sort of use cases like, you know, what can you gain?

And that can feed into, you know, traffic sort of things, or it can feed into agriculture sort of things. But, um, something that encapsulates the GIS formats and, you know, understands the different layering, like, uh, a satellite with different resolutions, how do you align pictures from all of those? So there are different formats for that.

And, um, I think there's a lot of development needed on that, on those applications from the model side too. So I think that is that we are gonna see those things come, uh, you know, there'll be more like more dedicated companies first even figuring out the models to operate on these sort of images. I mean, in the past when we looked at medical imaging, um, formats, we essentially would slice them up.

Like, DICOM is like a series of, uh, p and g files, the usual image of format. So we would slice it up because DICOM itself contained too much. So that is like, you know, um, but that's when you become very domain specific.

Yeah, I'm glad that you brought up, uh, medical, because I think that that's definitely an area where we're gonna see a lot of development in this, but also as, as you talked about a lot of geographical data, um, I was talking to somebody who's working on drone technology and they are working with everything from, uh, you know, GPS data streams to topographic information like you mentioned to, uh, you know, real time feeds, uh, from sensors, and all of these things have to be integrated and localized and plotted together. It was a really interesting conversation a little bit beyond me, but, um, but I could understand the challenges because there, you know, it's not just video, it's not just maps, it's not just text, it's all of these things as well as lidar and radar and, you know, cameras and, and all of this had to be integrated. So I, I think that increasingly that's what, what the challenge is gonna be is how do we integrate all of this data in a way that a, um, an AI agent can understand and act on without just overwhelming it with data?

I, I think you bring up a great point, and, and you know, it, like anything in, in, in AI right now, it's a two part thing. You know, there's the model, it needs to start having an understanding of it. Um, and, you know, there are the, the multimodal models are definitely, uh, you know, um, advancing rapidly, and there is that data part.

So, you know, one of the unique aspects, so why did we bring in a graph into picture Originally it wasn't because we were thinking there's gonna be all this knowledge graph use cases and things like that. We brought it in because it gave us a good way to represent relationships, and it was flexible to let us represent whatever data type we wanted to represent in it. So in our same graph structure, and we use a property graph structure for that reason, instead of the, um, RDF um, graph that come in that just like, you know, sub we don't do the subject predicate object representation, we do the full, like that if there is a representation for people, it'll be like a person node in the graph, you know, name, last name and all that stuff.

But it'll be, it can be very easily connected to another node that's a picture of that person, or that's connected to like video clips of that person and all of these special data types videos. Um, you know, we can introduce lidars documents, all these things have representation in the graphs. You can go from one type to another, and in the same query you can be saying, I want all of the various data things associated with Stephen and Frederick together, like wherever they appeared together, whether there was a text description, whether there was an event, whether there was, uh, you know, recordings.

It, it can go, it can use the power of graph reversal to get there. Um, so that's why we kind of, you know, originally started with the graph and, you know, now you can basically represent a lot of, uh, application information in it too. Yeah, I think one of the problems too is that, uh, not only is there a large amount of data, but the, the amount of metadata associated with the data is also getting more complex.

Right. So you talked a little, a little bit about the medical and geographical, I mean, the amount of metadata surrounding it is, is creating an additional, uh, problem in the complexity of the model. So, so, so one of the questions I, I had for you was, how do you see Agen AI evolved in the next 12 to 18 months?

I mean, if you look at MCP servers, they are less than maybe around a year old. It's going so fast. What's, what's your vision for AgTech AI in the next 12 to 18 months?

I think there's gonna be a lot more focus on what does it mean to get agent agents in production. Um, you know, we've built a lot of toy agents, we've built a lot of, uh, like, you know, agents that are starting to do some serious work. Uh, but I think especially in, uh, larger companies, you know, now it's time to go from POCs into production, which means really answer all these questions.

So all that we discussed, you know, how does, how does it get them maximum ROI you have to start thinking about your stack. Like, are you gonna do a framework way? What framework is the best?

What sort of models give you the least amount of hallucination and get you the most distance in terms of, uh, you know, your particular use case? So like, you know, we work a lot in retail and e-commerce, and that is personalized recommendations. Sometimes it, that doesn't require you to be a hundred percent precise.

You know, you're recommending product, you're telling them what you can buy. It's okay if like one of the products you recommended doesn't exactly fall in in that umbrella, but we also work with some medical copilot use cases, and there it becomes very important that you do not hallucinate. So the guardrails become really important.

So there'll be a lot more increased understanding in terms of, okay, for the vertical that you are in, um, what are you okay accepting and what are you not? And then what does it mean if you wanna go in production, what are all the data types you're gonna have to involve? What teams have to come together to put this information?

What are the guardrails that are gonna be, how are we gonna evaluate? How are we gonna observe and monitor this stuff? How are we gonna capture user preferences at, at scale without disturbing their experience?

Um, I feel like there's gonna be a lot more, uh, you know, uh, focused and organized efforts. And so the tool like, you know, platforms like ours, uh, become really, uh, key in, in making that happen. Um, I do wish though there is also some effort around taming compute.

We've been throwing so much power, and you can see these numbers about, you know, the electricity consumption for AI applications as like literally been drying reservoirs in places because of cooling. Um, I really hope there is some effort around that too, to reduce the energy consumption. You know, it's interesting.

I was just gonna say, it's almost like people need some kind of special database that can handle all this multifold data and maybe a platform that could bring it all together. Um, yeah, it, it is, uh, I, I think what people need to know is they need to know that such a, that such technology exists and that it is possible to bring together various data types and with AI applications and that, you know, people are working on this because I wonder how many people are just, you know, sort of dismissing it outta hand and saying like, we just can't handle this, or we don't know how to handle this. Um, so I, I guess, um, what do you see happening next, uh, from the industry overall in terms of integrating multimodal data with, uh, agentic ai?

I think it's gonna, I, I think it's gonna increase at a much more rapid pace, uh, with the, I mean, you know, there is at anytime the big cloud, uh, big companies start talking so heavily about it, you know, they start talking like six to nine months early because they're trying to build up hype around it. But if you went to, um, uh, Nvidia GTC earlier this year, or Google next, or like, you know, reinvent late last year, multimodal was already the thing and agents were already the thing and it was naturally like multimodal AI agents, right? Um, but of course the practicality follows a little bit behind in all of this.

So, um, yeah, so, uh, I, I think we'll see a lot faster adoption, especially like, you know, we are in, we are in production, so, you know, people can really unlock the data part and the moment you unlock data, um, the computer is ready. Alright, well thank you so much for this. It's been, it's been a very thought provoking as was, you know, our previous conversation.

And I hope that our listeners are starting to say, wait a second, maybe it's not, you know, just about text and just about structured data and, you know, passing JSON between, you know, agents and things like that. Maybe it's, maybe it's more than that. And hopefully that's the sort of thing that can come from this season of, uh, utilizing tech where we're gonna be talking to a bunch of folks who are doing some really cool things with AI agents.

Um, before we go, um, please, uh, let us know where can we connect with you, where can our listeners connect with you? Where can they learn more and where can they con continue the conversation? Yeah, so I am very active on LinkedIn, so please connect with me on LinkedIn.

I suppose you'll share the profile, um, as part of the description. io. Uh, and I would say give it a try.

The cloud has free trials, so if you sign up on cloud, do aperture data io. Um, you can try out the database, you can try out our various workflows that make it really easy to ingest existing, you know, data example, run some embeds, try out the ui, everything is there. And if you are concerned about, uh, privacy because you know, you work at a company that won't let you send data to a SaaS tool, then we also have, um, free community edition on Docker hub, and you can definitely try all the database features, uh, through that as well.

And we would really like to grow our community. We have a Slack channel, uh, and we really, you know, uh, amplify people who build and contribute, uh, to the set of applications that can help end users. Um, so for sure, looking forward to such contributions and more multimodal agents built on top of adb.

Yeah, I can't wait to see what people build. Um, and Frederick, how about you? Yeah, I'm also active on LinkedIn.

com websites. And you will see both of us at AI Field Day, uh, which is coming up real soon here at the end of October. Uh, we're pretty excited to, uh, be bringing together a cool group of companies, uh, talking about various, uh, elements and aspects of ai, some of whom you will hear about on this episode, or this, I'm sorry, on this season of, of utilizing tech and, uh, hopefully some of whom, uh, we will connect with further.

Uh, if you are excited about AI and, uh, agentic AI and, and where this is all going, uh, do check out the Tech Field Day website. com. Uh, that's the website.

Um, techstrong AI is our media site and also, uh, we're gonna be launching another podcast, a weekly podcast focus on AI as well. So keep an eye out for that. So thank you you so much for joining us and listening to this episode of Utilizing Tech.

Uh, you'll find this podcast in your favorite podcast application, just search for utilizing tech. You'll also find us on YouTube. If you enjoyed this discussion, we'd love to hear from you.

Please give us a rating. Please give us a review. Uh, please subscribe.

Uh, this podcast is brought to you by Tech Field Day, which is part of the Futurum Group. com. Or connect with us on X Twitter, uh, blue sky, mastodon, or, uh, yeah, LinkedIn.

Uh, you can look for utilizing tech. Thanks for listening, and we will see you next week. Okay.