Techstrong TV July 3, 2025

Hey, everyone. Happy July 4th. Let Freedom Ring Freedom.

You're watching Text On Gain. Hi everyone, it's Alan Chival here for Tech Strung Gang. Hey, day before July 4th, it's our last show of the week.

We won't be on tomorrow as we celebrate freedom in its many forms. And, you know, freedom could be like a, like a Jonathan Livingston seagull thing. If you, if you let it go and it comes back to you, it's yours.

If it doesn't, it never was. And it's not something, you know, from, uh, many of us, we, we've taken it for granted living in, let's call 'em Western democracies, uh, personal freedoms and freedoms to do many things. And, you know, this is a year where I think we need to reflect and say, uh, what's that freedom worth to me?

Right? Am I willing to give up freedoms? But anyway, we'll talk more about it.

We've got that and more. We've got some healthcare cyber news on here and some GI ops news and more. We've got a great panel to talk to you about it with.

Let me introduce you. They've been on before. So you know, these folks out west.

We've, we've got the, uh, the czar of the Valley, John Swartz up in Canada, our Canadian DevOps leader, Garima b Powell, although she's based in Canada, to be fair. Ri Garima's, a worldwide DevOps leader, actually a big CD foundation, uh, leader as well. And then of course we have the dean in New York, Mike Ard.

Welcome, welcome guys. So Mike, it's another July 4th. Uh, it's rain.

It's supposed to rain down here. I was kind of bummed out. I wanted to go see some fireworks tomorrow night.

But really, July 4th isn't just about the fireworks, is it? Well, I hope not. There's more to it.

But, you know, it's interesting, you have this essay that's on Textron ai and I encourage everybody to go read it. And at least I feel like we're kinda on the, on the cusp here of having to reexamine the social contract by which we all kinda live. And, um, some of the question is as to what degree is AI going to just be something we turn loose or do we need to come to terms with some way to kind of reconcile or manage this thing in a way that, uh, respects everybody else's rights?

And what's the definition of that? And John, of course, has had a piece talking about how there was an effort to, uh, limit the state's ability to regulate AI that was abandoned in this recent big, beautiful bill at the last moment. But, um, you know, walk us through your reasoning here.

Sure. So you, you mentioned the social contract, Mike, and, and there is, right, if I was a political science history major, and, you know, we learned a lot about Rousseau and Locke and Voltaire and some of the other Philis Phil philosophers that our founding fathers here in the US kinda looked at, uh, Montague in, in how, in how they fashioned our social contract, right? We have, we have some principles in the US that, well, we had some principles in the us things like your rights end at the tip of my nose, right?

You could do what you want as long as it doesn't infringe on my rights. Things like better. 10 guilty men go free than one innocent man goes, goes to jail.

Um, personal freedoms, freedoms around economics, around, you know, the, one of the great things about the United States, and quite frankly, you know, modern Western civilization is economic mobility. We, and created more middle class people than the world's ever seen. Um, but every, every, you know, there are boundary points and there are forks in the road that we come to and some bigger than others.

And it seems like AI is one of these things that has the potential to be bigger. The internet, certainly as big as anything since the internet. 'cause when you think about it, the internet has wrought all kinds of change to our freedoms and good and bad ways.

We all thought it was great that everybody can have a soapbox to go off and talk about and give their opinions. But we forgot that opinions are opinions and facts are facts. And we're still dealing with the repercussions of that.

Now, AI promises to have maybe even a bigger effect than the internet on our, what we do, how we do it on our freedoms. Uh, you know, they always say about Singapore, for instance, Singapore has had great stability, security, prosperity, and they probably have traded off some personal freedoms. And the people who were okay with that, it was their social contract.

Places like China, you know, where we have, uh, less personal freedoms perhaps, right? And the, and the government, uh, is more involved in our day-to-day activities. But, you know, we're past the point of pointing to China as the bad example.

And we're somehow that still that shining city on the hill that Ronald Reagan spoke about. Well, I don't know if that city is quite as shiny when it comes to talking about our freedoms. And AI has the potential to make it a lot less shiny.

But here's the beauty of it. AI also has the potential to make it brilliant, brilliantly shiny things it could do to equal the playing field. To give everyone an opportunity to bring more, to do more.

Is there, I think it's up to us and not our government, us each one of you watching this, it's up to us to figure out what's the right use for ai. How do we all leverage ai? What can we do to make ai make our lives better?

Not take our jobs away, not increase the police state. You know, with, we've spoken about many of these topics over the last couple weeks and months, but what can we do to make our lives better using AI to let freedom ring? And that, that's the really the point of my essay that I wrote.

I do encourage you to watch it, but I think this is gonna be a pivotal year, a year to three years where we gotta figure this out. 'cause it could go either way, right? It's, I feel like we're at midfield and I'm flipping the coin.

And John, you could call heads or tails. Mm-hmm. Yeah, I mean, we're early.

I mean, we're early in the narrative about ai and this one thing about this technology that I haven't really seen from many others is how polarizing it can be. Like there are stark opinions or, uh, points of view, whether it's, it can be incredibly good or incredibly bad. I, I think like everything else, it will end up a mixture of both.

And I think for the most part, it's gonna be positive. I think our healthcare will be better, transportation's gonna get better content, democratization of, of certain people and things. But the always, the issue I think with this, as well as with the internet and previous technologies, is that people need to be responsible in their use of it.

And I think as, as we've seen, they, they, they try to figure out a way or an in on, on, um, profiting from it, whether through illegal activities or just the land barrens that we have right now. So it's gonna develop, I mean, it's, I think you're right, LAN I think the next two to three years we're gonna really have a good definition of how this is work works within our, our homes or, or our, our workplaces. I think it's gonna be predominantly a positive experience, but right now we're kind of in the, um, the, the, the, uh, the boogeyman phase where it can do so many things badly or it can do so many things incredibly well.

So we'll see. See, Is this gonna be a social contract that we all agree to? Or am I someday just gonna go up on some website somewhere and train a bunch of AI agents to defend my inalienable rights at the expense of everybody else, and everybody else will do the same, and then we'll just fight it out on the web or in ether somewhere.

I think that's a good point. I mean, I think, I think we kind of do that right now, and I think social media ramped that up. So we're already accustomed to it.

Maybe it intensifies with ai and that's kind of terrifying because I think of what you could do with ai, um, much far farther beyond anything that you could do with the internet or Facebook or X or what have you. I think of deep fakes, I think of like these influences, not just in our personal lives, but our, our own personal expense in terms of scams. Um, it's gonna be like a wild west.

Yeah. And I think we're gonna probably need some sort of regulation, which kind of brings me to this big beautiful bill act, which would've basically done away with any type of state laws or state regulation for a decade. Thank God that that went by the wayside.

I I mean, ironically, even the most conservative Republicans, like Marsha Blackburn wouldn't stand for it, um, even Ted to, to Ted Crable people. But it, it, you're right, Mike. I mean, this is, this is a danger we are going to face and it will happen.

It's just the extent of what's gonna happen and how farther it may perhaps divides us or brings us together. You know, Mike, you mentioned in alienable rights, and we take for granted what we think those inalienable rights, right? Our creator hasn't doubted us with inalienable rights among them, you know, life, liberty and the pursuit of happiness.

I think I got that just about right. Um, I think we're at a point where we've gotta decide, really, what are those inalienable rights? I mean, if I could be stopped by some mass men in the street who, who claim to be, and maybe are in fact federal agents, but don't have any idea, refuse to give you ID, and they got masks and guns.

Well, I thought we had inalienable rights around that, right? We have a bill of rights, illegal search and seizures, stuff like this, habeas corpus things that our system has been built on. If we're gonna start trading those or, or cheapening those and let AI infringe on those inalienable rights, well then they're not inalienable anymore, right?

And that, that's the key. Johnny, you point about the big beautiful bill. I'll point out, it's not quite signed yet and it's not quite done.

It was kicked back to, had the house. We don't know if this is gonna stay in or not. But, but here's the thing about it.

What it really was is it precluded, precluded the individual states from doing anything on ai, right? It was, it carved out for the federal government. And under the current administration, federal government mean that that could be, you know, again, back to the inalienable rights.

That could be a scary thing. Now, there's a little bit more history though. It wasn't as clean as, as just, oh, we're gonna take it out.

Initially, the Senate said, well, there's a little too crazy, just a flat out carving out of no state action for 10 years. So the initial version of the bill said, well, no, the states can, can legislate around ai, but if they do, they forfeit their rights to this 500 billion, 500 billion, let me say it again, 500 billion pot of money that if you're going to use it for broadband or ai, you forfeit your right to that $500 billion if you're gonna legislate around ai, which was, I guess, a little better than just flat out pro prohibiting them from doing it. Right.

A little bit more of a carrot than a stick, you know? And, and traditionally states like Mississippi, Louisiana, Alabama, they don't care about the money. They don't spend it on that anyway.

But even as you say, John Marsha Blackburn, who I don't think I've ever agreed with, one thing she's ever said publicly spoke up and said, wait a second, here's what she really said on the floor of the Senate. The federal government has a terrible record, a terrible record of, of legislating this stuff. Where's the real action taking place?

At the state level? It's the states who are consistently here in the us. 'cause we don't have the political will nationally to get things done.

It's the states who've come out with cybersecurity regulations, privacy regulations, trying to draw the line on where our inalienable rights cannot be trampled on, and to prohibit them from doing it. Because we think somehow the federal government's going to do it is really having no AI legislation at all. And we need to keep the states involved.

Now, whether the house will buy that or not, John, I don't know. Yeah, I mean, we're gonna, I mean, because of the, when we're we're filming this, it's probably gonna happen. It happened or didn't happen on Wednesday, but who knows.

But the interesting thing about Blackburn, you're right, Alan, I I have exactly the same opinion. I probably don't agree with her on 99% of what she, her thoughts are, but I have interviewed her and I found her actually to be a voice of reason in terms of what the federal government doesn't do, because the federal government has been absolutely hopeless, including anything like online privacy, AI related legislation. I, I think it'll be la say Fair any 20 years.

Yeah. I mean, 25, uh, Copa, it's like the last major or only significant, uh, related technology regulation they've done involving privacy or, or the internet. So, um, you know, it's good, it's interesting.

It's not, it's really a very weird bipartisan viewpoints from extremely liberal, extremely conservative folks tend to be in agreement on some of these things. So it's really hard to pin down. But, um, you know, the influence too, of Sam Altman and, you know, he testified in May about this, uh, about latest about regulation of, of ai, and he made a very strong argument against it, obviously, because his company is part of, uh, Stargate, the $500 billion AI infrastructure movement, which is kind of plotting along.

Um, that scares me. That's actually what scares me is the influence of some of these folks, especially in the tech side and their influence on, on government and, and the, the Trump administration. So That's, you Guys realize the rest of the world is kinda looking at the, this whole conversation in scratching its head.

Because in places like Canada, for instance, the assumption is, is that if the authority is not expressly given in the provinces, it resides in the hands of the federal government. Whereas we're the kind of the only country that has the opposite of that, and everybody else around the world is going, what are you guys debating? But, but here's the thing, and Garima, you may or may not be up on it, I don't know.

Canada's actually already enunciating some regulation around AI and so forth Privacy. I'll provide a technologist view on this because, you know, from a technology standpoint, I see there is a rise in a new role or a new era for tech diplomacy, because a lot of these issues are kind of, you know, uh, related to the demography, the people sentiment. And, you know, the tech diplomacy is needed in certain aspects to ensure that there is a balance in how innovation can drive, you know, uh, you know, your healthcare forward, your digital infrastructure forward, but at the same time does not hurt, you know, the community sentiments, you know, protect the rights, uh, the privacy rights, uh, building data ethics and the frameworks.

So I think there's a lot needs to be done before we enforce, uh, these policies and decisions around, uh, a wider or broader community. So from a tech perspective, I believe that there is a lot of, uh, tech diplomacy needed in this. And, you know, even in Canada, we have different provinces, different needs, different needs of digital infrastructure.

So, I mean, there's no straightforward answer to this, but yes, again, the question is, you know, when you talk about democratization, who is accountable? You know something when something goes bad, right? Yep.

So that, that's the question to be answered. And of course, the eu the eus out ahead on this, right? The EU iss already spoken and, and acting, aren't it?

But to your point, Mike and, and John, about, you know, the $500 billion, um, project, what's it, spotlight? Searchlight, Stargate. Stargate SG one, right?

Stargate. What, what's interesting here, right? It recently came out, nothing's been done with Stargate yet, and nothing's been done.

There was a, in typical fashion of, of the administration, we had, we had bright lights, big city press conference, made all kinds of announcements. Nothing's been done. All of these pledges and everything else sound great.

In the meantime, the government's gonna put a $500 billion fund into ai. So 500 billion of, you know, of government money perhaps going in there. You know, let's not lose sight that this big beautiful bill conservatively adds three, three and a half trillion dollars to our already high deficit, three and a half trillion dollars to the deficit that will bring our debt to like 125% of our GDP.

And it's, it's done on the backs of, of taking out money from people's Medicaid and, and college financing and other things to give billionaires and millionaires great tax credit so they can invest in AI and things like this. Guys, I'm gonna end, we gotta end this segment, but I'm gonna end it with this. It's July 4th.

People died and fought for your freedom. Don't stand by and let them take your freedoms away from you. That's all I'm gonna say on it.

We're gonna take a break. We'll be back. You're watching Textron Gang, Discover Techron 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.

Hey, folks, we're back and we're continuing this conversation about politics and while also adding a little criminal element to it as well. 6 billion healthcare fraud scam. It's been going on for a while and involved all kinds of folks from nurses to doctors to, you know, full blown criminals.

And John, I gotta tell you, here's the thing that drives me crazy about this. 6 billion going out the back door somewhere. And I'm sure there's more of this, and I don't really mean to see and sound like, you know, I'm anti-government, but this is crazy.

I mean, for all the Doge investment and for all the stuff that we've been doing, can we just figure out how to maybe contain the fraud so that we have enough money to support the people we need to support? Yeah. Yeah.

I'm afraid that we we're gonna probably see more of this, uh, uh, because as we all know, when, uh, new te tech technology comes along or something that is existing is, is amplified, like the internet under ai, we're gonna see a lot of bad guys take advantage of the system. And here they definitely did. Um, there was this absolutely complex fraudulent schemes where these foreign operators, I, I think 30, nearly 30 people in Estonian Pakistan were part of this, uh, this transnational criminal organizations, um, were accused of buying medical supply companies to which they file false claims.

One of the things that, that jumped out at me, and it kind of refers to our first segment, was that executives with marketing companies in Pakistan had allegedly used deepfake technology to create these fake recordings of Medicare patients who apparently were agreeing to receive certain healthcare products and then filing false claims. So there's a use, again, of, of ai. Um, at the same time, our government doesn't want us to do anything about that for now.

Um, again, as you said, Mike, um, criminals with the bad guys really do take advantage of, of technology because they, in a sense, are like the pioneers in its use in many ways. And then we try to counteract what they do. And, and the great irony, as you said, is Doge was all about efficiency.

Governments, uh, lack of getting rid of waste. Yet here we're seeing this incredible amount of waste that of all things involves Medicare, uh, at a time when we're talking about, about cutting those benefits for folks as part of a big, beautiful bill. So it's all kind of a crazy world, but it's not entirely surprising.

And my fear is that, um, healthcare, which I think has the most potential in terms of AI use, will also be a target. So it's gonna greatly benefit us down the road in terms of how we look at healthcare through ai, but we're also gonna probably encounter more schemes and scams and potential dangers to our, our, not just our health, but our pocketbook. Yeah, I, I'll, I'll jump in if it's okay.

By all means, I have more to say, but go ahead. How, how many people out of these 324 are gonna serve in Congress of the Senate one day? Because we, we've got our fair share serving right now who have been implicated, or their companies have been implicated, convicted, and had to pay for Medicare fraud.

Medicare fraud seems to be rampant. This is the late, it is a big one, 14 point a half billion, but, you know, you can't throw the baby out with the bath water. But, you know, for all the noise about Doge and everything, and all these, you know, big cost cutters, seems like, you know, some work needs to be done in that system.

We gotta design a better mouse chop. Yeah, and that comes back to my next point, which I know we have this notion in the land of cybersecurity that we don't blame the victim. But you know what, I think in this case, there's some folks who work for these agencies that need to get fired.

It's clearly somebody's asleep at the wheel, and at this tune of 14 billion plus, God knows how much else is going out there. It's broken. So we need to kinda like really have a conversation regardless of whether you're a Democrat or a Republican, this system is fundamentally broken, needs to be rebuilt from the ground up.

And maybe we need a commission to go do that. That's bipartisan. And this is all they do.

We, we don't do bipartisan commissions anymore. That that's an issue. I think occasionally we can come up with an issue that maybe we can all agree on it.

This is just insane. Well, the, the problem is you're going to get one group that says the whole thing's broken. Scrap it.

You're going to get another group that says, well, we can't scrap it because it's mission critical. People's lives depend on it. And there's gotta be a middle ground where, no, you're not gonna scrap it, but you're gonna make a pe.

You know, this is the kind of thing, and I'm a child of my times, right? I, I get this, but this is exactly the kind of thing where Bill Clinton would take Al Gore and Al Gore would lead that bipartisan commission, and they'd come up, maybe it'd take 18 months, and it wouldn't be on the headlines every day, but they'd come up with a plan for Medicare two, or whatever you want to call it, that would, would, would do things like this, right? We don't have those, you know, Lyndon Johnson was given a charter to build a space program, and we wound up in the moon in eight years, not because it's easy, because it's hard, right?

Clinton gave Gore to reinvent government, and for the first time, and the only time in my lifetime, we actually ran budget surpluses. When was the last time we did anything like that? That's what's needed here.

And it's hurting folks in these red states more than it does. It's where it's pretty clear that a lot of those folks are more dependent upon i i These services. This isn't a red or blue issue.

The fact that it's wrt with, you know, potential for fraud is, I don't think anyone argues. I think that the question is, is that a reason not to do it all together? Or what do you do to fix it?

What do you do to fix it? I mean, let, let's face it, we're, we're the last industrialized country that doesn't have socialized medicine anyway. And all these other countries figure out a way to do it.

And in spite of what the media may tell you, they do it well, they do it well, we can't figure it out. And, and we get this craziness. I mean, people, people complain about those services in Europe all the time, but last time I, They complain about our services too.

But, you know, at the end of the day, the standard of healthcare in Europe and Canada and other places where they have socialized medicine is pretty damn good. You know, I, the last time I was in Singapore, I, I, I cracked a tooth and I had to get a root canal, some of the best dental care, best equipment I ever had. I think it cost me, I don't know, $65 probably cost me a thousand dollars or more here.

Right? That's with insurance, right? With insurance, exactly.

That's with my insurance. So, you know, this nonsense that people complain, everybody complains, everybody. That's how, that's the nature of that beast.

But I'm telling you that there's gotta be a better way. I think it's like the whole climate too, is like, we used to have a lot of bipartisan bills, like you mentioned Clinton and LBJ and all the legislation, especially with LBJ that was accomplished, and now it's, there's so much divisiveness. Um, we, going back to Clinton's administration and LBJ are talking about pragmatic, very smart people who knew how to get things done versus rank amateurs who are not especially brights.

Yeah. Which is the climate we're in now, or have been in now for, um, it will be a, oh, But even, I mean, look, I don't want to just pick our democratic administrations. Ronald Reagan got things done, Reagan, I, I didn't agree with a lot of his positions, but he got things done bipartisan, poppy Bush did, right?

Well, Reagan was successful because Tip O'Neill was willing to have the conversation. Well, it takes two to tango, certainly it takes two. And, and, you know, look, I see what goes on in New York.

Not gonna get all political, but you see what goes on in New York City with this mayoral campaign and stuff, and, you know, can you blame people? Who should I vote for here? The guy who left the governor's mansion in disgrace, or this other guy who's a Democrat socialist, maybe anti-Semitic to boot who, you know, you're not giving them a choice.

All right, let me, let me vote for the, for the criminal. Who's in, who's in there now anyway, right? Th that's not a choice for people.

This isn't just a Republican issue. The, the, the, the blue states and the Democrats are just as guilty for failing to give people a, a viable choice. Oh, yeah.

It's, it's a choice. That's an issue for both parties. I mean, they can't work together.

I mean, it's, it's not, you don't blame one side that we blame both sides. It's, you know, and is it any wonder that no one wants these jobs? Right?

Right. I can even jump in here. Go ahead.

Right. So let's steer this away from political to technology mess, right? So I think from a technology standpoint, uh, I spot two big problems in this.

One is the scale, and the other one is sophistication. So we have to tackle both of them, right? And when you look at like all these compliance checklist, uh, HIPAAs of the world and high techs of the world, I think this is a wake up call for them.

You know, try to kind of also bring some money on the table. Try to kind of introduce new sophisticated audits, checklist assessments, because the, the fraud at this scale, if it happened, I mean, technology can be a lever to put the positive side of all this, right? So how do you, um, control the supply chain, for example, what security controls you have in that I identity access management, if somebody is using defects, right?

So all this can also bring some kind of positive, you know, revolution or evolution for healthcare in the technology side. And I, I strongly advocate that somebody needs to look at the audit assessment and the, uh, HIPAAs of the world to see how we can improve this in the best of, you know, the Technology. So I, I agree with you, Garima.

This is a, at the end of the day, a technology issue. This, this particular, you know, Medicare, there's no reason, and maybe that's what a panel would need to do, put in design a better system, design a better system, do Not let a good crisis go in waste. Come on.

Alright, let's take a break. Let's get off the politics. I promise you.

We're back. Next. We're giddy up GI ups.

It's something safe. You're watching. com is the leading resource for news analysis and education on challenges facing the cybersecurity industry.

com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more. com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more. com to learn more.

com. Home of Security Bloggers Network. Hey, folks, we're back and Alan promised you a political free discussion, but I think he just lied, but we'll get back to you in a second.

com, put together by the folks at Okta Post Deploy. And it's suggested people are adopting GI Ops alongside new platforms. So therefore, it's evolutionary, and it's mainly being driven by the rise of things like Kubernetes, and then people are willing to reconsider their software engineering practices.

There's also a whole conversation going on around platform engineering, but Karima, what's your take on what's going on here as GI GitHubs and platform engineering kind of, uh, joined to the hip somehow, orthogonal to each other, and what makes people actually change their workflows? Okay, so let's reflect on the findings of the report first, and I was looking at the report, uh, it seems that, uh, 660 software development professionals, uh, put their views onto this report. 93% said they have adopted Git ops in some way, shape, and form, right?

And the maturity of that depends on different factors. There is also like an increase in adoption of GitHubs. Uh, so there is, uh, a definite kind of rise of, you know, interest in the community.

And GI tops has helped, you know, industry segments to kind of push the needle for reliability, um, auditability collaboration and all those kind of, uh, you know, good components which, uh, technology can bring forward. Traceability is another one, right? So I think, uh, GitHubs has come a long way.

I wouldn't say that GitHubs is a silver bullet though, because, you know, it has its own flaws. If you ask the community, I mean, the open source community is very vocal about things which doesn't work, or GitHubs doesn't fix, right? So if you think about like large organizations with, you know, multiple repositories, siloed environments, there is a lot of challenge in onboarding to such kind of, you know, revolutionary change, right?

So if you have seen the report, 7% also say that they would stop using GI Ops. I think they, these 7% work for large ecosystem or large enterprises and probably are, uh, stifled by the siloed environment and the ecosystem. And there are many environments, uh, in like big enterprises, right?

So there, they're, where I think platform engineering comes, uh, as savior. And I, I also feel that, you know, platform engineering will not be a server bullet. Um, as you say, I mean, it works for big enterprises, to be honest.

I mean, platform engineering, if you see like how, and who are the takers of the, this kind of movement is like big large enterprises, tier one customers, right? So those are people who will be, uh, you know, onboarding to that. And, and of course, uh, if you think about the challenges which we have like spoken about from a GitHubs perspective, platform engineering could resolve those challenges in certain ex extent, and they can bring some advanced tooling, which I can talk a later, a little bit later.

I, I also want, uh, other panelists to be contributing to this discussion. So I give it back to Mike, you, and maybe you can come back to me, uh, for some more future roadmap items. Absolutely.

Alan, I can't help but wonder sometimes when I look at all this stuff is, is this just like fashion trends and software engineering? We just kinda like bounce back and forth between things and, you know, this, this, this month, you know, it's wide lapels and, and thin tots. You know, I don't want to sound like one of these old fogies who say, oh, there's nothing new under the sun, but there is nothing new.

No, I'm kidding. There are new things. But look, just a couple words here.

First of all, this survey comes to us from Octopus Deploy. The people who acquired, uh, Codefresh, who are the people behind Argo, which is the main GI ops, uh, platform, open source platform, and there's, and they offer hosted GI ups. Um, so, you know, you gotta take the findings, I think, with a bit of a grain of salt.

And so I, I would caution and, and mention that. Secondly, though, you know, I was at Platform Con last week in, in the in-person day in New York City. The virtual event went on all week, and there was an in-person day in London.

You know, they used to say about platform engineering. Well, once you got your Kubernetes set up, it's all good, right? That that was the beginning and the end of platform engineering.

But no, it's, that's not, it's not just Kubernetes. The big emphasis in platform engineering now is on what we call IDP, right? Internal development, internal developer platforms.

And you are right, gerima, it's a big company game, right? Because you've gotta have enough developers to go develop a platform or use a platform that they're all going to use. So you have some standardization.

Now, part of that IDP extends into the DevOps CICD tail, if you will. And that's where the GI ops lies. So this is right in the sweet spot for platform engineering.

You are using Kubernetes, you got an IDP, why not incorporate, you know, Git ups in there as your way to, to move that along. GI Ops has always been closely linked with, with Kubernetes, with Cloud Native, right? Because the old CI CDs Jenkins, some of the older ones like that, didn't really work with Kubernetes and cloud native natively.

Um, gis, GI ops did. So I, I do think in, in spite of the obvious bias of, of the company doing the survey, I think the results are pretty much show what the, where the market is, it is being adopted. I think larger organizations that are using IDPs are using GI ops as part of that, and they, because they increasingly are using a cloud native architecture, microservices, and everything that goes with it.

Yeah, I will actually add to this discussion because, you know, uh, whether it is platform engineering or, uh, GitHubs, you know, there's, so, like I start with GitHubs, like what key challenges, which are not addressed by GitHubs. And then we can also discuss if they are, uh, addressed by platform engineering or not. So, for example, the big challenge which we see from a practitioner's perspective is secure, uh, secrets management, for example, GitHub's has been struggling with that, uh, inherently, and it doesn't solve that problem of secure secret management, right?

So probably this is something which needs to be solved through GitHub's community itself, you know, the future roadmap item, and, you know, how do you integrate, uh, many, uh, you know, environments and storing secrets and get, uh, you know, and making it safe, right? Um, another, uh, main problems when you see the adoption percentage is also related to branching and environment management. So, to a certain extent, environment management can, like the problem can be solved by platform engineering, right?

But branching strategy is another issue, which probably platform engineering will not solve, but probably the community and the, the teams and the enterprise themselves have to think about a structure, how to create, like how to not to create an unnecessary complexity in the environment, right? So these are, uh, few things and challenges. And if you think about, uh, portability or vendor lock in, maybe platform engineering, uh, is a good kind of savior in that context, because probably, uh, with GitHub's, uh, if, uh, this is a concern from the community that, you know, it's limiting portability or it's tight decoupling tools, I think platform engineering can also become some, some kind of, you know, um, change trigger in this, uh, visibility.

And auditability, auditability, for example, is a big, uh, area because, you know, it's not very easy to relate what changes happened when, you know, live deployments happen. It's, it's still kind of, uh, you know, in the inception stages of, you know, bringing that control and traceability with GI tops. So hopefully the, this will be a future roadmap item and the GitHubs community, for example, where, uh, you know, we can have an AI assisted GI ops tooling to resolve these kind of problems.

And if you remember, uh, our, like, last discussion, Ellen, around MCP, right? So the flux, uh, community has come up with an NCP server, for example, which integrates that, that capability and brings agent tech AI into the mix. So that's another area of, you know, um, looking at it from a GitHubs perspective, how do you make AI technology more accessible for GitHubs?

And, you know, what kind of ecosystem, uh, you know, uh, connections you can make. So we can be more ecosystem centric. We have to have more, uh, flexibility in the approach how we are deploying this into a large ecosystem or large enterprise.

I gi I leave you with these, uh, points because I think this will also give some indications of the work needed to be done in the community for GI tops. You know, Alan, we talk about these topics a lot, and there's always a conversation about all the things that need to be improved in the land of software engineering. And yet, every morning I wake up and there's just tons and tons of applications that seem to be deployed working.

So like, what, you know, is this just magic? 'cause if you go to a business executive, he goes, what problem? The softwares, Well, you are right.

You know, we are, we are generating, whether it be through AI or not, we generate more code and more applications every year than we did the year before by a wide number. No doubt about it. But it's also because software and technology is becoming so much more important and ingrained into almost everything we do.

So you need, you need more software, you need more applications. And you know, we, we forget the whole idea of writing, you know, software and, and creating applications is what, 75 years old, right? It's not that old.

And, and so it's still to a, it it's still more art than science, though. I, I think we've made a lot of progress towards science, making it a science, not just an art. You know, it used to be developers basically crafted code and applications in like a, a vacuum, like a, like an old time craftsman in Europe working on wood or clocks or Swiss, you know, Swiss folks making watches.

Where now it's, it's much more industrialized, it's much more scalable. It's much more, it's not that craftsman per se. Hey, we all could be software developers in the world of ai, right?

Um, so we're, you know, the, there's so much riding on our software. Every company's a software company. Software is eating the world.

There's so much at stake, of course, it's, it's continuous improvement, continuous Involvement. Yeah. You, I mean, well, it's, uh, also not, uh, shy away from the fact that no tool, no platform can save a poorly designed software Yeah.

That this is true too. Crap in and scrap out. So how much of our software is poorly de designed?

A lot of it just outta curiosity, A lot of it, and it can always be made better. That's, that's the nature of this beast. Mm-hmm.

But we're getting, so here's the good news. We're getting better every day, and we're, we're, you know, the science gets better, the the best practices get better. And, and that's, you know, that's why we sit here and talk about these things.

Otherwise, we'd be talking politics all day. But if I'm a developer and I perceive myself to be an artisan, am I gonna really, you know, emotionally warm up to the notion of science? And we talked about this in previous shows, but who the hell wants to go to work in a factory in the first place?

Well, this is true, this is true, but I think, I think the generation of software developers who think of themselves as artisans, Look, I I, I would say that, you know, as a software practitioner, I have the left hand side of the brain and the right hand side of the brain, right? So the left hand side of the brain is co-creation with new technology, more creative side of it. And this is, again, community driven collaboration.

And right hand side is more systematic, system oriented system architecture. All this like, is required as an individual. So I think both sides have to grow together.

Agree. My take my left brain and right brain argue with each other all day long, which is why you find myself seeming mumbling to myself down the hallway. Yeah.

Well, but, but that's what it takes. It takes, it takes a whole brain. Anyway, hey, we're over time.

I got to, uh, end this one. Um, John Reemer. Mike, thanks for joining today.

Thank you for watching this episode of Textron Gang. We do have a full, uh, uh, Textron TV schedule immediately following, so stay tuned for that. If you're watching this on the live stream, if you're watching this on, I don't know, Textron tv or the YouTube channel, or the OTT, thank you for, for taking time to click in.

We appreciate it. As I mentioned, we will not be on, uh, Friday, tomorrow, Friday, uh, July 4th for, uh, independence Day here in the US and wherever you are, cherish your freedoms. Enjoy the weekend, and until Monday, then This is Alan Schmo for Text Drunk Gang, wear Out.

Hi, everyone, welcome back to Tech Trunk tv. My next guest is Tanya. I hope I get this name right.

I practiced. My next guest is Tanya Obra. Obra?

No, OBRA. Devic. Obra.

Say it for me, Tanya. So, my name is Tanya Ovi. Ovi Ovi Obra.

You say it so much better than I do. I gotta tell you the truth. When you say, Oh, you're doing pretty well, you're doing Pretty well.

No, I know, but you say it with with feeling as If I know what I'm talking about. Yeah, yeah. Well, me, you know, I'm just, you know, I'm like an ai, I'm a hollow, I'm a hollow impersonation of a real person, perhaps.

Oh. But in any event, Tanya is the senior manager of Java programs at the Eclipse Foundation. Tanya, welcome to Text Drunk tv.

It's great to have you on. Thank you. Likewise.

It's very nice being here. And thanks for the invite to Alan. My pleasure.

So Tanya, how, how did you come to be the senior manager for Java programs over at the Clipse Foundation? I know it's been a long journey, actually. I am, uh, involved with the Java for a very long time.

I don't wanna say, uh, how many years, but let's say, uh, Java is 30 years old, so I'm there from the get go. So really, there is my connection with, with Java. Java.

Now, I was there, um, uh, working with Java in various capacity. Um, I was a developer. Uh, I was doing consulting, a course developer, you know, um, uh, doing program management for various companies.

I worked for Oracle for many, many, many years. And at some point in time, um, uh, Oracle has decided to contribute the whole Java EE body of work to Eclipse Foundation. And, uh, me being involved in Java, and particularly within, uh, with the, uh, enterprise Java, for most of my career, this seemed to be like a good fit.

So that's how I ended up, uh, being at the Eclipse Foundation. So from the, from the get go of Jakarta ee, if you will. Um, and then, uh, uh, my responsibilities grew a little bit more, um, and, uh, um, covering the team that is, uh, um, working with Java related projects, and particularly, uh, in, in, in the formation of, uh, the, uh, indu various industry collaborations.

Uh, and that's how I ended up, uh, being, uh, uh, lead for Java programs at Eclipse Foundation. I love it. What a great, what a great journey.

Good for you. Hard to believe 30 years, right? I, I know, I remember, I remember, I remember Crazy, crazy times.

You know, I started a web design company in web, we became web hosting 1995, maybe it was so 30 years ago. Yeah, 96. Yeah.

We went to a client and they wanted to have a spinning globe on their website. And back then there wasn't anything. There was no Java script, there was no, there was nothing.

There was Java. And my partner designed a transparent spinning globe in Java. It was beautiful.

Of course, this was when we still had modems, right? So, you know, if you didn't mind waiting two to three minutes for it to download it, to start spinning, it was beautiful. Anyway, um, Tanya, you know, I, I've been involved.

I'm older than you. I've been involved in tech space a long time, long time. And you know, I, I look at the way open source is today versus the, when I first got involved in open source, it was Dr.

Richard Stallman and what I call the cathedral and the bazaar. If you ever read that book, kind of the wild west of, of open source. Then we had sort of this big brother era of open source where Sun Microsystems, for instance, would be the, the, the caretakers for Java and just Sun.

What they decided, they decided, right? They were good kid. I don't want to give it a bad context.

They were good care keepers in, of course, it got sold to Oracle and, and everything. But, um, but now we're in what I call the foundational era of open source, and we have the Eclipse Foundation and the Linux Foundation, and the Apache Foundation. And the beautiful thing about these foundations, it, it allows competitors to cooperate.

So coopetition for the good of everyone, and it's a great, and, and, and as a result, open source has never been more powerful, more popular and more widely used than it is. So thank you to, to the Eclipse Foundation, and for all the work there, it's why we're able to do these great things that we do. Um, you mentioned Jakarta.

Not everyone in our audience is gonna know Jakarta. How would you describe it to people? So, uh, as I mentioned, um, uh, uh, Jakarta EE is a su, uh, successor of Java ee.

So yes, basically it is the continuation of the technology that existed for, um, quite a few decades now, uh, as well. Um, and, uh, it, it came out again, it's, it's great how you tied the open source into the whole conversation. So, um, we, uh, Java came as an open source project from, uh, one vendor into, uh, eclipse Foundation, where we're taking really, um, uh, uh, care and making sure that we are preserving vendor neutrality.

So, uh, meaning you don't have the dominance of one vendor deciding on what is going to happen in the technology. And, uh, that actually created a major shift, um, uh, uh, uh, for, for, uh, Java E. Um, and if you, if you really look at it, uh, Java, e for, for, um, various reasons, uh, uh, in the technology in different, uh, um, uh, pushes that were happening at the time, uh, with Oracle was a little bit neglected.

Uh, we, we were all aware of it. So therefore, it was the community and, and everyone who was, uh, dependent on this technology, who kind of, uh, um, uh, pushed a little bit more, uh, for the idea of, uh, uh, uh, contributing the body of work of Java E to Eclipse Foundation. And once that agreement was put in place, and we started, um, uh, receiving the, the, the technology at the Eclipse Foundation community, again, uh, raised their voice and decided to rename the technology to mark this shift into Jakarta E.

So basically, the last release of Java EE was Java E eight. And, uh, um, uh, the first release to honor the continuity of Jakarta EEE is Jakarta E eight as well. So, um, first initial, um, couple of releases, uh, were really dedicated to making sure that we, um, uh, properly, um, uh, have the base of this technology moved to Eclipse Foundation, um, to have a different, uh, uh, process for releasing of these, uh, uh, of the work, particularly of the specifications.

And then, um, uh, so defining the process, uh, um, you know, rebranding, um, uh, the governance model, uh, making sure that we establish the vendor neutral, uh, way of developing specifications. Um, and then with, with, I would say, um, um, uh, uh, Jakarta release nine, there was a major shift that was happening, uh, with adopting the new Jakarta, uh, namespace. We had the restrictions on the use of JX namespace, so therefore, we had to do this major shift, which was, I now, uh, looking in hindsight, I'm looking at, uh, as a, as a blessing, uh, in disguise, almost like, yes, it was a major work, major disruption, uh, uh, uh, in the industry, but it really showcase to what degree technologies out there, um, are dependent on the, um, uh, uh, Jakarta.

And, um, uh, one by one, um, you know, we, we heard, oh, we need to deal with the namespace change. We need to, and now we're supporting, uh, Jakarta nine and, and so on. So, um, uh, uh, it really, um, uh, alerted the industry to, uh, to Jakarta now.

Then we, we released, and that was a, um, timeframe of, uh, 2020, uh, within two years, we, we released, uh, Jakarta 10, which was 2022. And, and now we're releasing Jakarta 11, and I'm calling the timeframe is, uh, 24, 25. Um, uh, and, and, and, and we're keeping in this, uh, two year-ish type of, uh, cadence.

Um, uh, I will talk about why two year-ish and why 2024 and 2025 in a minute. But just wanna, um, uh, uh, highlight that. Um, we are, uh, already anticipating 20, uh, in 2026, um, uh, Jakarta 12 release.

So, um, this story is really showcasing the interest of the community to, uh, for the longevity of this technology. And it, uh, actually shows the, the interest. And, uh, you will see as we start talking for a specific release, uh, for this specific release, that, um, uh, there is innovation also coming into play.

There, there is a lot of work and a lot of investment from the existing community and a lot of work that is actually, um, uh, enabling inter people interested in this, uh, body of work to easier, um, uh, uh, become part of the community, um, uh, with the use of the newer technologies and not being stuck in, in, in, in, uh, time with this technology. Um, so, uh, a lot of I agree. Yeah, Yeah.

You know, Tanya, sometimes you, sometimes you gotta make a clean cut. Yes, right? Because otherwise it could just fester and draw out, and you never, so as painful as it might have been back when you did nine, it was probably the right thing to do.

Secondly, you know what, I think it was Shakespeare who said, right, arose by any other name, would still smell as sweet. And, uh, no matter what you call it, it, the usefulness of it, and, and its, and its use in industry and the in the world is evident, let's focus in on Jakarta EE 11. Yep.

What's, what's the big story around this release? So, the big story really about this release, we call it, uh, a release, um, about performance and develop developer productivity. Um, those are the, like every release, we have some keywords.

This is the, this is what we're, um, uh, uh, how we describe this particular release. So, um, the, the, the, the, the key elements of this release is one brand new, uh, specification. 0.

Then were, um, also, we did a major, major work on the, um, uh, test compatibility kit, test compatibility kit for, for, um, uh, Jakarta is crucial because what we are doing is, um, we are all the implementers of these, of, of Jakarta technology, um, uh, have the opportunity to be, uh, a certified, um, uh, on, uh, as, as Jakarta EE product. So we are having this test that every implementation, uh, needs to go through in order to claim compatibility and use Jakarta logo, for example, to showcase, uh, um, their, um, compliance with, with the, uh, specifications. So, believe it or not, uh, uh, the, the, uh, test, uh, compatibility kit was not, um, revamped or has not been modernized, uh, or restructured since, uh, the day it, it was initiated, which was, uh, quite a couple of decades ago.

Uh, so it was using old technology and, uh, um, it, it was, uh, limiting, um, entry barriers for, for new developers to, um, innovate, uh, in, in the tests as well as it was, it was, uh, kind of limiting new implementers to go through the certification process. So that's, um, uh, the, the team, uh, the community, um, uh, involved in, in, in Jakarta 11 actually took time this year. Uh, and that's why, um, I was referring to 24, 25 ish, uh, uh, timeframe, uh, because the, the team really took time to modernize the, the, the test compatibility kit.

And we're now the tester base on the Uni five and, uh, Apache Maven. And, uh, it was restructured from this monolithic, um, uh, uh, body of work into, um, uh, uh, uh, different modules that are, uh, uh, possibly specific to one specification only. Um, and, and we can easier build up on, on the, on the, um, new tests to add, add new tests to the, uh, TC case.

Um, so we, just to illustrate the amount of work that was necessary to be put into this, um, uh, the, the, the old individual specifications. So the Jakarta is, is, is, is comprised of over, uh, 30, uh, individual specifications. Um, all the individual specifications were ready for release mid, uh, 20, 24.

However, the team really, uh, cares about, uh, uh, the quality of the releases and quality of the certifications and, and, uh, compatibility program, um, uh, that they decide, okay, well, let's step back, uh, put the time necessary into modernizing these tests. So, uh, we can, um, uh, uh, see the benefits later on. And basically, if you look at it, they didn't anticipate it will take almost a full year.

It's not almost, it took a full year. Um, uh, but, uh, uh, we as a consequence decided, or, or the community rather, has decided to, uh, start releasing in phases. So as, as you, as you, uh, may know, uh, Jakarta ee, uh, consists of, of, uh, a couple of profiles and then the platform, uh, uh, that includes all of the specifications.

So, uh, the smallest grouping is, uh, uh, Jakarta core profile. And, uh, that was ready for release in December, 2024, and it was released. So Jakarta, uh, E Core 11 was released in December, 2024.

Then, um, uh, we released, uh, in March, 2025, uh, Jakarta 11, uh, web profile. And, uh, now finally, we are on June 26th announcing the release of the Jakarta, uh, platform. In the meanwhile, we also have the, the, uh, uh, compatibility, uh, products being, um, uh, showcased on our website.

So we do have already, uh, um, uh, core, uh, profile implementations, uh, uh, compatible with the Jakarta 11, same with the web profile. And we're expecting, uh, very shortly that we will see, uh, quite a few, uh, adopters of Jakarta 11. And, and beside those two major, uh, pieces of work I would like to call out that we are continuously, um, uh, improving many other specifications.

So pruning what's not being used, um, updating with a little bit, uh, uh, modern technology, uh, um, uh, we are, uh, continuing use of Java records, for example, uh, throughout the, the, the specifications, um, uh, removing references to security Manager, um, as, as Java se and, and j uh, four 11 is, is, is, uh, um, uh, requiring. So there is, there is a lot of, uh, um, uptaking, uh, that is, uh, uh, also, um, uh, uh, uh, taking place. Take place.

Yes. So, uh, we are also, uh, for 11, or that is, uh, mostly interested for any Java developer, um, doing the enterprise Java development. They always want to be on the latest, uh, uh, Java se.

So we are supporting, um, uh, and based everything on the Java SC 17, however, just because, uh, everything can run on on Java, uh, uh, 21 as well, uh, uh, Java 21, uh, features can be used throughout, uh, including virtual threads that, uh, many developers are interested in. So a lot of, a lot of, uh, good work, um, was happening in the 11, but I have more. Okay, Go ahead.

But wait, there's more. Go ahead. So, um, uh, the interesting thing, when, when, when the community realized that the TCK development is going to take quite a bit of time, they actually, uh, decided that to start the conversation about Jakarta 12 in parallel.

So the team that is focused on TCK uh, work is, is relatively small, and I absolutely have to call out the efforts that were, um, uh, done by, by Red Hat, uh, on, on, on the, on the, on the TCK work. Um, our, uh, release lead, uh, from Microsoft, uh, also pushed for the conversations, uh, that are, um, uh, Jakarta 12 related. So we are now having actually solid plans for Jakarta 12, and that's why we're, uh, kind of, uh, uh, calling it out that we're hoping to release Jakarta 12 in the, uh, the year 2026.

Um, so the, the, the, the, the, the, the release plans for individual specifications, um, uh, were already, uh, done for 12 and, uh, there are, uh, planning their plans for the core profile, uh, for web profile and the platform 12. Uh, uh, also, um, uh, uh, the, the release plans are, uh, uh, done. Um, so between the plans and realization, there is a long, um, uh, uh, uh, long time, uh, a long step to do, but nevertheless, the work has already started.

And, um, it is, it makes me, um, extremely proud, uh, to, uh, how we're doing in, in Jakarta world. I love it, Tanya, that that was, that was a lot you packed in there the short time. One thing you left out Yes.

For people who want to go see this, download it, read about it themselves, where do they go? Jakarta ee I love, it's, that's easy. Easy.

How simple is that? Yes, fantastic. Good for you.

Good for you guys. Yes, Tanya, congratulations on what sounds like a jam packed release here. You guys have put the time and effort into it.

It is a multi-year effort. Good luck with Jakarta 12. We'll have you come back on that.

And good luck to the Eclipse Foundation. Keep doing what you guys are doing. You make, you make open source what it is.

So thank you. Thank you for your efforts. Thank you so much, Alan, and it's great to be talking to you.

Hope to talk to you soon. Absolutely. Lemme see if I get, take one more shot at this.

Tanya. Obvi, obvi f I'm getting there. All right.

Yes, Jakarta, EE 11 from the Eclipse Foundation here on Tech Drunk tv. We're gonna take a break now. We'll be right back.

Hey, everybody, we're at the Open Source summit in Denver, and we're gonna have a little chat about PyTorch with Matt White, who's the executive director for the PyTorch Foundation. Matt, welcome to the show. How you Doing?

Hey, very well, very well, thank you. PyTorch, the foundation is expanding its mission apparently, and that was some of the conversations that were in the keynote today. So beyond PyTorch itself, what, what are your ambitions for the foundation?

Sure. Yeah. So the foundation's actually grown quite, quite considerably since it started, uh, in 2022 with, um, the Linox Foundation.

So prior to that it was with, um, meta. And so in these last few years, and in particularly in the last about six months, we've grown to become an umbrella foundation, which means that we're able to take on additional projects beyond just the PyTorch framework. Um, and this puts us in a really good position to help steward be good stewards of open source AI and sort of advance the cause and, and provide tooling and open source that embraces open principles and, um, moves the dial for us.

And so some of these projects, uh, satisfy really important parts of the AI lifecycle. And so the ecosystem has grown, and we recent, recently announced that we added additional projects, uh, VLLM and deep Speed, and we have a few projects on the back burner that, uh, we'll be joining shortly as well. So how far does that go across the lifecycle?

Because there are open source AI coding tools. I mean, how, how ambitious can this whole program gap? Sure, sure.

So our, our objective really is to equip AI researchers and engineers, and basically scientists and anyone else even hobbyists to, with the tools that they need to be able to do either research or productionalize workloads or even just learn about PyTorch and learn about AI and, and deep learning. And so as far as we're going on, the horizontal is really from data ingestion, so data pre-processing and storage all the way through for, you know, model training and then right to the, you know, tail side of that, which is like inference and serving. And so that really gets us through this like horizontal, where we're also building up vertically, including like AI frameworks, agentic frameworks, and all of the tools that are needed by mainly enterprise to build applications on top of AI models as well.

So does PI Torch the foundation become the primary vehicle for open source AI development? Or will there be other foundations alongside it that need to kind of collaborate and what will be the relationship? Yeah, so I think even within the like specter of like Linux Foundation itself, there are other adjacent projects and as adjacent foundations that we, you know, have collaborations with.

And if you look at like the tech stack for, um, you know, let's just say like, like serving side for ai, so serving models to, um, end users on, you know, the very bottom of that. You have the Linux operating system and the kernel and built on that you have Kubernetes and you kind of keep going up. And so there's always gonna be this like, interplay of different foundations working together.

Um, and so we do collaborate with other foundations on AI directed initiatives as well. Yeah, When all of this first started out, there was no shortage of tools and it was kinda, you know, to a lot of folks it kinda looked like gobbledygook, right? There was all kinds of different things to do, all kinds of different things.

Is that starting to coalesce a little bit? Or there certain tools that are maybe becoming defacto standards that are kind of more widely adopted? And have we reached some level of maturity there?

Yeah, I think the, I think the industry has sort of spoke at least on the like training side with PyTorch, which, you know, hugging faces announcement that they're going to be strictly based with PyTorch and they've, you know, decommissioned jacks and, uh, TensorFlow, uh, from their, from their platform. So, and most contemporary models are built on PyTorch. Uh, most distributions of open models are in, uh, you know, PyTorch ready format.

And so I think from the, at least from the training side and model redistribution side, uh, you know, PyTorch has being sort of the, the defacto, um, you know, uh, framework of choice on the serving side. There's still some, you know, discussions, but VLLM has kind of skyrocketed up in popularity and, uh, you know, they joined us, uh, earlier this year as one as our initial, um, you know, umbrella foundation project. Yeah.

And VLLM is an open source alternative to Cuda and whatnot, and those types of frameworks. 'cause I don't think everybody knows what VL Yeah, Sure, sure. So VLM takes care of like loading a model and serving it and, and doing all the inference.

So when you go on, you know, Chachi pt a query, a model that on the side on the back end of that, it's serving. And so that's what VLM solves, right? Is being able to serve up tokens or serve up bits or whatever the case may be, whatever, um, modality it is.

And so it's, you know, torch Serve was a, a project that in the PyTorch ecosystem that did serving, uh, there's others out there like SG Lang. And so, but the industry's really rallied around VLM and a lot of folks like IBM, red Hat, you know, meta Cisco, others have kind of put their weight behind, um, this serving framework. And the Idea is not to get locked in on the inference side to a particular platform, Right?

Right. Yeah. And is that gonna apply not just the GPUs, but other classes of processors?

It's kind of, I mean, how Open is open, yeah. So like VLM team, just like the PyTorch team and others that are working closer to, to the metal, um, work with those vendors, right? And so work with a MD Nvidia, work with Intel, you know, Qualcomm arm, so forth to make sure that models are performant on their particular silicon, right?

And so this is always like a constant process of evaluating, and as you know, PyTorch evolves as some of these other platforms that work more closely with the Silicon evolve, they work sort of hand in hand, um, to make sure that models are performance on these, uh, these different, uh, systems. Coming back to, uh, PyTorch, um, it's, you know, it has a relationship with Python, which is widely adopted by folks beyond just developers. So will we get to a point where we'll see more so-called citizen developers working with PyTorch to go build and train models?

I mean, how expansive can the target audience get? Yeah, I think you, there's some limitations with like, especially large models, right? Like we're, when we're talking about smaller models, like for recommender systems or these classifiers that they often use for, you know, identifying whether content is, you know, like for guardrails, like identifying whether content is, you know, has, you know, foul words in it or something like this, right?

Um, and so these, these are a little more accessible because there's a smaller, uh, sort of like hardware burden to being able to train those models. But when we talk about large language models, there's significant hardware involved. And so I think where I'm seeing we're, we're sort of seeing the most, like usage is actually on the fine tuning side.

So someone builds a foundation model, you know, large organization or a lab builds that foundation model, releases it with a permissive license, and then downstream users take that model and fine tune it on data for their particular application and create sort of like something that's very tailored to them, or they, um, experiment with it. And, you know, there's a lot of hobbyists out there that can actually never really need to touch PyTorch because they have all these like, web accessible tools, right? Like hugging face, uh, for example, right?

And there's some UNL and some other platforms that really make fine tuning and other, you know, quantization and these other sort of transformations with models much more accessible. And so I think going back to like, you know, PyTorch, it requires a certain set of skills, you know, to your point on, on Python. Um, but also having access to the infrastructure needed to be able to train at scale Has been been a lot of debate about what open source means in the context of ai.

And some people are saying that their models are open and other people say, well, they're not as open as they should be. 'cause they don't have, like, the weights are not included, or I didn't understand how the data was set up. How should we be thinking about open source in the age of ai?

Yeah, That's a pretty, um, that's a good question. Uh, it, it's, you know, there's different folks that have different beliefs about what this, you know, when they take like the open source paradigm and apply it into world of AI where there's, you know, in the, in the conventional space in open source we have software period, right? Um, now we have all these other artifacts that go into the mix of creating a model and creating some, you know, output that can be used by downstream users, right?

And so there's, you know, training code, there's inference codes, there's different types of data sets that are used for pre-training or fine tuning and so forth. And then obviously documentation, research papers, there's a lot of things that can go into that. And when Linux Foundation published a, um, a something it's called the model Openness framework about a year ago or maybe a year and a half ago, which set forth to kind of look at what, what the definition of open science would be, what would be an open model in this new paradigm, right?

And, and when we're talking about openness, and so we sort of went through these two dimensions of like openness and completeness where the industry had sort of settled on this term openness to kind of cover the licensing cover, how many components are released. We went and looked at, okay, completeness is really like how many components are released. And then openness is, is it released under an open source or open license, right?

And so we try to disentangle that to make it much more clear to the public, like what each is responsible for. And I think conflating those two things makes it very difficult to understand like what open source AI really is and what open models are. But if, you know, within our, our, um, classification system, an open model to us was, you know, a model and its associated weights, but released under a permissive license, right?

And this is what gives you what sort of embraces that spirit of openness because an open source, it's about being able to use modify study for any purpose without limitations. And so we were able to replicate that with, with models through that framework. Most of the usage of models, at least the ones we hear about anyway, seem to be tied up around proprietary ones, is the open source models gonna become more dominant as we go along.

'cause history would show us that open source seems to eventually catch up and then it becomes the innovation engine when everybody starts contributing. Yeah. Yeah.

And we see this time and time again, right, where things are sort of released in as black boxes are developed in house, and then their open source community replicates them, and then at when the sort of performance differential is so minute, eventually, like folks really rally around the open solutions and then the innovation happens on top, right? And so with, with models, we're seeing that the, like, you know, there were big leaps in open source is way behind black box solutions for models. And then we sort of saw that gap close, and now we're at the point where a lot less companies are releasing these foundation models because the differential isn't so extreme, um, between, you know, the last generation and this generation.

And I think we'll start to see less innovation potentially on the model side, although it's not going away, um, but a lot more interest in innovation on the system side, which is now where we're seeing like people that are looking at like, how do I integrate with this back office system? How do I get data out of this particular database? And, you know, we're seeing the rise of, uh, protocols and standards like MCP and A to A and others to try and solve a lot of these problems.

But now that we're more systems focused, we're gonna see a lot more innovation in that space and, you know, continue to grow around innovating on top of these foundation models that may not, um, you know, they'll continue to incrementally get better, but the, you know, the industry may not be as model focused as they are systems focused when we try to, you know, operationalize these platforms. Yeah. Last question.

Sure. Um, and I hear this all the time. Everybody kind of, you know, they hear open source and they nod their head and they have a warm feeling for it, but they don't know where to get started.

They don't know how to join the community and they don't seem to have a good handle on where do I get involved. So if you want folks to, you know, come work with you on the PyTorch Foundation, where should they get started? You know, where, where's the interface?

Sure, sure. Yeah, I think there's, there's a few entry points, right? org, which is our discussion boards where people can glean a lot of information from that.

We're also actually building out, um, you know, we do currently have like tutorials and other artifacts online, but we actually have two programs. The foundation is launching, uh, this summer, which are focused on training and certification. And the other is with, uh, what we call like academic outreach or osbo outreach.

So we're building the packet so that educators can teach their students how to get started with PyTorch, how to get off the ground with it. And then on the training side, we're actually building training materials and certifications for both entry level and advanced, uh, you know, PyTorch skills. And then we also have like our Discord server and Slack, which are good forums for people to kind of get involved and start asking some of the questions about like, how do I get started?

And, and, uh, you know, how can I train a model and experiment and these sort of things. All right folks, you heard it here. There's a new epicenter starting to emerge around open source ai and it's at the PyTorch Foundation, so check it out.

Matt, thanks for coming by. Thank you very much. Appreciate it.

All Right. Yeah. And we'll be back in a minute.

Hey everybody, welcome back to the open source summit in Denver. And we're talking with Clyde Siad, who's general manager of Linux training for the Linux Foundation or training in general across the board. And they've got a new report out talking about what is the current state of tech talent.

And so give us some of the highlights of that, some of the things that you've found, and you've been doing this for a while, so what surprised you? Yeah, sure. First off, thanks for having me, Mike.

Always great to be able to get on platform and talk a little bit about the work that we're doing at LF Education. We've been doing this for several years. It used to be called the, uh, open source jobs report.

And a few years ago we thought, you know, every job is an open source job in tech. So let's just broaden the lens a little bit. And you know, this year of course, we focused in a little bit more keenly on the AI topic, which is sucking all the oxygen out the room.

And we wanted to understand from practitioners, if you back away from the hype, what's really on their mind? And, you know, it turns out what's really on their mind is this realization that I think the stat is like 96 or 97% of organizations see that there's potential significant value in ai. A lot of them are realizing that you only to get that value unlocked, you have to go through your people and you have to cross train and upskill your people because you can't insource, you can't outsource your AI strategy 'cause it's fundamentally about business processes.

And so I think this, the impact that has is there's been a lot of stuff in the popular press around consume for jobs, entry level developers. And you know, one are the impacts gonna be, and that's true, but if you take a bigger lens picture of what's gonna happen and how the sort of deck chairs get rearranged, there are gonna be significant pockets where there's gonna be the need for, need for more technical talent at the people level in order for organizations to unlock the potential value. And organizations are realizing and saying, I think it's two ing that they need to invest more in, in technical talent in order to unlock ai.

So, and I think we, we, we, uh, titled the report about something about the vibe versus the, the reality and, and part of the reality that the data is saying is as that first wave of enthusiasm and maybe panic has, has sort of washed over the real work of what is it gonna take to get there is coming into focus. And a lot of that's gonna have to be using the folks who we have who understand the organization, the processes, the customers, to figure out how we add these tools in to drive productivity. You mentioned new jobs, new skills, new areas.

Are there any that are becoming apparent to you where we might have new roles and new opportunities for people in the age of ai? Yeah, I think it's more the intersection of rules, right? And so we already know from the DevOps revolution that it broke down the wall that we used to have between the dev dev side of the house and the engineering side of the house.

I think what we're seeing with AI is it's breaking down the wall between the tech side of the house and the business side of the house. So if you think about what it takes to build a really good agent, it's not all about the technology. It's very deeply about what the process is and what the outcome is we're trying to achieve.

And so I suspect what we're gonna see, yes, maybe fewer entry level jobs for software developer, because a lot of that stuff is, is quite good with the agent coding and maybe more opportunities of people who come in with some technical talent, but also are upskill on the business so that they understand how are we using these tools? What are these we using these tools to accomplish? So I, I think the, the early read is there's gonna be more of these, um, I think the Tim is like T-shaped folks, right?

They're deep in one area, but they understand across the business because that understanding across the business is gonna be important. 'cause you need that to unlock the power of what you can get with the, with the agent models. So I think the nature of the rules, right, they're becoming these broader rules because you're gonna need access to that broader set of things instead of, like historically we've gotten super comfortable, right?

I mean, you remember there was a time when you could train to be a database person and then do that for 30 years and then hang it up and leave and, and we've known that that's been gone for a while. Mm-hmm. I think what we're seeing now is you kinda have to not just get across the different technical functions, but you have to start reaching out across into the business and really understanding that very deeply.

And yet I still see DBAs out there. So who knows? Listen, there are still QA folks out there, so, you know, it's, uh, it's a long tail To your point though, not just the way our jobs are gonna change, but the organization itself may change as as well.

'cause if the, all these silos become, um, fungible then, you know, the separation of roles between say sales and marketing and customer service and all these things that we've structured the companies around forever, might that also change in a way that the new job functions will be cross disciplined, to your point. And the way the organization is structured needs to change too. I suspect.

That's right. And, and I was saying this in a talk yesterday. We in an uncomfortable time right now because we all had playbooks that we were pretty comfortable with and we understood how they work.

And you came in at this level and you progressed to that level and there was career tracks and that's kind of gotten set on fire. And now we're all trying to collectively figure out what did the new playbooks look like in this sort of cross-functional T-shaped world. And I think the answer is we don't know.

And the fact that we don't know is creating a lot of anxiety, which is natural. I think part of the coaching we're trying to give folks is take a breath. It's okay to be anxious.

We're figuring out the new playbook. We have a sense that it's gonna be this multidisciplinary approach. I don't think anybody has figured out exactly what that looks like, but it's not, it's not the approach we were using three years ago, right?

It's gonna have to be something quite different when we think about how do we bring the power, you know, uh, the context about the business is not captured in any models and, but it's gonna be critical to unlocking the power of the models, right? Mm-hmm. Mm-hmm.

So that new, the new playbooks are still very much being written. Have you seen any generational divide here in how we're thinking about ai? Because I mean, I can talk to folks in their twenties and some of them will say, well, this is great.

AI will empower me to do all kinds of stuff that I previously would've had to have 10 years of experience to do. Others in that same age bracket are saying, um, I'll never get a job in this field 'cause I can't break in. 'cause all the entry level tasks are now being automated and, and they feel a little bit stuck.

On the other hand, the older folks are standing around going ranging from, you know, this is a conspiracy to eliminate my job to, um, this is gonna be great because now I don't have to hire a bunch of entry level minions to do stuff. I can just have the AI agent do that for me. And in, in their minds it will benefit them more.

Yeah, I think it, that is absolutely what we're seeing. I'll start on the first one. I do think there is a significant divide among the entry level folks coming in in terms of their tolerance for ambiguity.

The folks who can deal with ambiguity and uh, figure out kinda where they're gonna skate to and what they're passionate about, are the first group of people you're talking about that are seeing it as opportunities. I think the folks who are anxious in the face of uncertainty because they want somebody to tell them exactly what to do are gonna have a lot of trouble. We are entering an age now where you have to be flexible, you have to be nimble.

And it's frustrating for young professionals because it used to be that you could get in and you can build up a certain level of competence and expertise and that then felt like a safe cushion to sit on. The reality is by the, if a slight overstatement, by the time you get to be an expert on something, it's no, it's probably no longer relevant. It's just moving so fast.

And so this idea of like never being comfortable in your mastery is a very different way of kind of existing in your career. And I, I, I think you and I are maybe further along so we probably won't have as much runaway to deal with it, but I tell my kids who are teenagers, you know, this is what's gonna be, you're gonna, you're gonna have to just be constantly evolving your skillset. And if you can wrap your head around that, you're gonna do fine.

If you are panicked by that, it's gonna be tough sweating. My, my son was on a job interview recently and he was talking about the job and I asked him, while you're worried about, you know, what AI might do to that particular field? And he, and he looked at me and he rolled his eyes and he said, dad, if I ain't gonna worry about AI all the time, I'll never get out of bed.

So, Right. He's On the right track. Right.

And then I think from the, you know, mid-career and later professionals, there's also this disconnect between the folks that feel like they've earned their stripes. You know, they came up through, uh, there may be slightly annoyed at just how good the models are at at sort of digesting and analyzing the sort of state of knowledge. Uh, I suspect some of them are a little bit over indexed on, I'll never need to hire an entry level person again.

Uh, I just had some, a plumber out to my house at $125 an hour. The guy is 62 and he said he'll never retire. 'cause he has more work than he can ever possibly do because there are no 20-year-old plumbers, right?

If, if you break the on ramp of talent, like it's great for a while and then it stops, stops being great for the customers and it starts being great for the, for the folks who are left, We, we might have a robot for that jump. Yeah. Well, not sure how I feel about letting a robot mess with my plumbing, but that's a whole other question.

Uh, but I think everybody's adjusting, right? And I think it's the, the folks who are in the middle of organizations are the ones caught in the pincher, right? Because the board members are like, oh, this is great.

You should triple your productivity. Just, just use the ai. Of course there's no one thing that is ai.

And then you've got the folks at the entry level. Some are panicked and some are excited. And the folks at the, you know, in the middle of the org are the ones have to sort of bridge the gap between the two.

Yeah. Because I do feel like there's a certain amount of noise coming from the C-suite execs about, you know, we're going to improve productivity, we're gonna reduce head count and all this other stuff. And, and as I listen to them, I can't help but wonder most of 'em are gonna wind up hiring people back.

'cause they don't really understand exactly how the thing that they're working through actually works Gonna play out. You know, I saw a great stat the other day, uh, in the US in the 60 years following World War ii, productivity increased on average by two and a half percent a year. 4%.

If you remember the hype back then turns out the internet made us less productive. We, we somehow lost ground. I think the AI tools have the potential to kinda get that productivity number back up.

Uh, if done right, the the trick is how do you do it? Right? Right.

Doing it right is gonna require leveraging everybody in your organization because it is fundamentally about looking at your company and figuring out how can we use these tools to make better decisions faster? How can we use these tools to get stuff out to our customers that is better mapped to what their needs are and have more feedback? Some of that's data, a lot of that is judgment and a lot of that is knowledge, right?

And so figuring out the people part of that, I think Boston Consulting Group had this great framework out that they're calling it 10, 20, 70 10 percent's about the tools, 20 percent's about the processes, 70 percent's about the people. We're super excited about the tools and everybody's talking about them. It's 10% of, of the challenge, right?

I mean I was having this conversation with one of my neighbors and he was pointing out, they had bought a manufacturing facility in Vietnam 10 years ago and there were 300 people working at it. And they have since tripled the output from that factory. But there's still 300 people working there.

And his point was we increased productivity, but we'd rather increase the productivity and the revenue. And then it's not just about reducing headcount per se, unless, you know, that may be a byproduct, but maybe not the primary goal. Well, You know, I don't think anybody can point to a company that ever shrank its way to greatness.

Mm-hmm. It's about growth, right? And it's about figuring out what more can we do.

So from your estimation, what is real in terms of the capabilities of AI copilots and now agents versus how much remains theoretical and what's the timeline spectrum in your mind? Ooh, well the timeline's a tricky one, but I will say what's real is I have a old friend who has long time, uh, con uh, contact center manager and they have significantly reduced their tier one staffing out of the Philippines because a lot of the questions that people call in to ask for the typical things that will go to a tier one desk are actually really good to put in model form because it's very well described transactions, right? I wanna reset my password, I forgot my login, I need a refund.

So there are all things like that where you've already seen pretty significant movement sort of in the real world. I think we're starting to see on the co-development side, uh, much faster velocities, right? Of people being able to, nobody enjoys the, uh, mundane part of the coding where you're stitching services together.

I almost nobody. So if you can accelerate through that bit of sort of assembling the blocks of code and you can spend more of your time thinking about what's the new functionality you're trying to create, I don't think you'll find a lot of people saying, oh no, I really like typing line by line. You know, all that code.

So I think that's real. I think the productive, I think the velocity argument is real. There may be some people who are feeling like they could be comfortable with largely model generated code.

I would argue most people are not. And, and they, and they still want sort of few review and guidance in that process. Uh, so there's something real aware.

I think the trick is figuring out what does that look like? Is it a velocity increase? It looks more like your factory example where you do more, uh, there some opportunities to pair back on kind of what the, what the on ramp of talent looks like.

The on ramp of talent doesn't go to zero because there's no business in the work and that can sustain that. So I think there's some basic stuff. The more interesting thing is when you get into the more you know the meat of the organizations, right?

Most things in organizations aren't as simple as reset my password. You're talking about dealing with suppliers, you're talking about dealing with customers, you're talking about dealing with multiple departments across your own organization. There's a lot of context and a lot of process and a lot of organizational dynamics at play that's gonna take real work.

And that's not work about, is it MCP or A two A? That's work about the horse training that happens within an organization as you reword processes. Uh, so it's not about the tools.

The tooth have a ton of potential, right? They're very good at inference, they're very good at analyzing yesterday's data. They're not good at envisioning the future because they're trained on yesterday's data.

So you know, if you go back 20 years, one of these tools probably wouldn't suggest the iPhone because there was no data suggesting that. You know, that sort of unique combination of factors, right? So innovation, a lot of the core innovation I think is the human spark layered on top of great analytics.

The analytics piece, the tools are exceptionally good at it. And to use your call center analogy, I'm not quite clear where those agents are gonna be 'cause I can see the call center people are gonna build their agents and they expect humans to interact with that. But I got news from them.

My agent's just gonna scan their site to find what I want to have in the first place and bypass their agents. It's gonna be very interesting to see how that one plays out. So, or maybe those two agents will talk to each other somehow or other and share something interesting.

Clearly everybody who's hiring anybody's looking for people who have AI skills. What's your best advice about how to go get those AI skills? 'cause I think everybody kind of nods their head and says yes AI skills but then they don't know where to put their arms around that.

It's a great question Mike. And it's something I've been saying for a couple of years now. Technical talent is something you should build, not buy.

I think a lot of hiring managers got into this mindset that I'll hop on LinkedIn, I'll find some candidates, I'll hire somebody and it'll be great. First off, that is a circular firing squad. 'cause I poached from you, you poached from him.

He poached from me and we haven't added any new talent. Secondly, it is logically impossible to hire somebody with three as a ENT AI model building. And so this desire, we, you know, we always want to hire the person that trained up on somebody else's dying.

You haven't seen that ad for 10 years of experience a AI agent. Yeah, exactly Right. So I think everybody's having is realizing that the easy way out, which is to hire somebody that learned it somewhere else is a logical impossibility.

And that leaves you with one choice. The people I have are gonna be the people that I have to rely on. And so how do I systematically invest in training and upskilling those people?

And you know, one of the things that was interesting in the research report was it wasn't just that people were saying AI skills, although that was the highest on the list. It was like some two foods, more than half of people said we're short on cybersecurity, we're short on cloud native basics. You know, we're short on DevOps and pipelines and the mechanics are sort of pushed stuff through.

And I think that's part of what we've been playing back to folks, is you can chase the hot sexy stuff to say, well we gotta do some A two A, we gotta do MCP. So we could build agents that talk to agents that manage other agents. Fundamentally this stuff is gonna run on the infrastructure we have, which is this sort of cloud native architecture.

It would behoove people to focus on making sure folks understand the basics. 'cause the abstractly are tools are gonna keep coming super fast. They're gonna keep running in our data centers where Linux is the command line and Kubernetes is the orchestration layer and Prometheus is doing logging.

You know, it's, it's a learn to read and write approach, right? Of saying, listen, you don't know what this is gonna be, but we know what this is gonna be. Make sure folks are comfortable here and make sure you're exposing them to all the new stuff that's coming out.

'cause it's every week, you know this like every single week more stuff comes out, but the base layer doesn't change. Yeah. Well we also know there's gonna be a lot more of this up here and this down here isn't gonna get that much bigger and better and Yeah, it's not, it's not sexy, but it's what we run on, right?

Right. So making sure everybody knows what you run on feels like a pretty good strategy if you're gonna bip the business on, you know, this this new family of, of technologies. So lemme ask you, who's responsible for training?

And I'm asking this question because uh, if I go talk to some of the attendees, they'll say they work for certain companies 'cause they give him access to training, right? Other companies expect the employee to kinda stay current and it's up to them and they to get that training somehow themselves. So how do you navigate those two extremes?

Because it seems like, uh, both parties expect the other to be doing something. You Know, my argument to organizations is if you think as 97% of people do that, there's gonna be this significant win from implementing these technologies. And you realize that the only way you're gonna get that win is by relying on the people within your organization to figure out how to correctly use these tools.

How to be privacy respecting how not to give away all your intellectual property, how to do it in a way that doesn't land with, you know, horrible backlash from from your customers. If you think that's gonna be what's required. But on the other hand you say, well they should figure it out themselves.

They're grown people. That's probably not gonna end well for you. Right?

Like, you'd never do that in any other aspect of your business. You'd say this is critically important, I wanna have a clear plan on a strategy and alignment on how I'm doing this. Upskilling your technical talent is no different than any other strategic initiative, right?

If you, if you think it's strategic, you have to act like it's strategic. And if companies don't and they instead say, well, but the people should be responsible for them themselves. I think what you're gonna start seeing is DV is the performance diverging between the organizations that take seriously the 10, 20 70 rule and the organizations that are still suspicious that if people train up, they're really training up.

'cause they're trying to leave to move to, to some other place, right? So if this is your strategy, you should align your strategy. And if your strategy requires people who understand the tech and are comfortable working in it and you, and you don't make that a priority across your organization to do it consistently, I don't think you can expect it to be as successful as those who make that commitment.

Alright folks, you're heard in here wise, man once said, failing the plan is planning to fail. Still true in the age of ai. Thanks for coming by.

Thanks So much, Mike. Alright. Hey everyone, it's Alan Shimel and we are live LIVE live in New York City at Platform Con in Person Day here in midtown Manhattan.

You know, platform con's been going on all week, right? So today's Thursday, it actually started Monday and like the last two platform cons, it's actually a virtual event and it is, I don't know, 35, 40,000 people registered for the virtual event. But what's interesting this year, they did it a little bit last year, but they really expanded this year are the in-person mm-hmm.

Uh, platform, Kanyes that kind of blend in to make a hybrid. So there's, there's one in London, which was earlier yesterday and of course now today they're in New York. Mm-hmm.

I am thrilled to introduce you to Preti Soma. Yeah. Yes.

Preti is, uh, SVP engineering at Temporal, one of the, uh, sponsors in exhibitors here today. And we're gonna talk a little bit about Platform Con and whatever else we want to talk about. So thanks for joining us Priti.

Thank you for joining us. Of course. Thank you for having me here.

And, uh, having platform com, uh, live in New York City. Yeah. 'cause I've always wanted to stay on camera.

Live New York. York Spice would say live from New York. Live From New York City.

It's Thursday afternoon. Correct. This is more exciting than anything else, right?

Exactly. Uh, very, I didn't think that's very good. Hello everyone.

And um, yeah, my name is Treaty. Uh, I have been in engineering for longer than I would admit and have been sort of on the platform journey throughout my career. So I started off, uh, at Oracle building platforms there long before they were called platforms.

Uh, spent some time at VMware, um, again, building platforms at Yahoo, which was a ton of fun. Yahoo was Yeah. In the days Yahoo was where it was happening.

Exactly. And, uh, at Yahoo, I was responsible for both building and running. And so that's where I think I was really hooked.

It was sort of my, um, the challenges that, that running platforms at scale with open source and all of the complexities that all the various Yahoo businesses had kind of really hooked me. Yeah. And, um, from Yahoo, I went to a tiny company called HashiCorp at that time, Absolutely no problem.

Um, I had used Packer, grat, and Terraform when I was at Yahoo. And, uh, I was fascinated by this platform shift happening again to sort of the infrastructure is code elements. And, um, landed in HashiCorp, spent five years there, uh, really building out the product.

I've been running engineering throughout. Um, and then went to temporal and as a, as sort of a fun fact, we were using temporal at HashiCorp as well. Really?

I didn't know that. So Part of our platform team at HashiCorp was using temporal and in fact, Terraform Cloud is also using temporal. Um, and that sort of, you know, temporal was just this really fascinating technology and, uh, kind of brought together a journey of both building and running platforms with like a platform product as well.

Love it. So how long have you been at Temporal then? I've been there for two years.

Okay. Yes. So you left HashiCorp about two years ago.

Was that bef That was before the IBM deal, obviously. Correct. I left about, uh, um, a year after the IPO took some time off and recharged and, uh, landed at Temporal.

Very cool. Right? Yeah, no, I've, yeah, I knew Mitchell.

com Right. Some of the other sites, we, we've followed the space for a long time. How would you describe Temporal to our audience?

Yeah, so temporal is essentially our mission is to shield developers from all of this complexity around running code that cannot fail. Now, that's a really loaded term, right? Code does fail, but what we do is we make those failures inconsequential.

And so what we have found is a developer, a platform engineer who is building something, is spending a lot of their time. In fact, some studies would say more than 50% of their time building kind of the, the scaffolding around all of the error conditions. And the magic of temporal is that we just do all of that for them and they just focus on their business logic.

You sounded, you would think a lot of people would like that. Yes, Absolutely. And we're seeing that we're an open source, uh, project.

Um, our business model is to deliver it as cloud hosted. And we've seen an incredible amount of adoption. Um, the product market fit is really tremendous, and our ethos has been to build this with the developer at the core of everything we do.

And so the attention to detail around the developer experience, uh, is just incredible. We, we build idiomatic SDKs, uh, we've got six or seven languages as of late. We are seeing Python sort of overtaking all of our other languages pretty quickly.

And, uh, we'll talk a bit about ai, I assume, uh, but you know, it's that focus on the developer that has really, uh, resonated and, and we're seeing a lot of adoption. So when you say you sell it as a hosted cloud, so is it like a, a SaaS multi-tenant hosted cloud service that you offer or you manage it on clouds for clients? Like I wanna run it on AWS We do temporal on AWS wanna run it on Google Cloud.

I do temporal on Google Cloud as well. Yeah. So we, we run the service, um, we run it and customers have the choice to create kind of, we, we, we provide a logical construct called a namespace where your mm-hmm.

Your workflow, um, sort of orchestration code runs, and that namespace customers can run that either in AWS or Google Cloud. One key thing I wanna mention is that our security model is incredibly elegant because what runs in cloud is, think about it is like the, the brains of the orchestration, the actual code, the SDK code, it runs in your own infrastructure. So we don't actually see any like Your own code.

Exactly. And so that's actually one of the reasons why you're Not, you've got the logic out here, but Right. The IP, if you will, is Exactly.

Exactly. And so the, the SDK I selling it exactly that your code essentially is just waiting for instructions from the cloud around tasks and steps and retries and like all of this logic that makes your code sort of failures be inconsequential. And this security model, the elegance of it, honestly, I think is one of the main reasons as well, you know, first developer focus and second security model that has really helped with the adoption.

It, It almost turns the, the classic cloud security conundrum on its head, right? Which is, I gotta trust this cloud provider to, to keep my stuff secure. Right?

Right. 'cause I only, you know, I only see it, it's like I only see the top of the iceberg, right? I don't see the rest of the iceberg under the water.

Right. And, and that traditionally has been a problem in cloud security, right? Absolutely.

So this is saying no, what, what's on the cloud, even if it did get broken into it, is just logic. Exactly. It's no, that's not the, it's not the ip, it's not the data that they, It's not the data.

And, uh, you know, the, the, the big thing it also does is because because customers, developers are outsourcing reliability, to us, it clearly means that the temporal system itself must be really, really reliable. Right? And so what we wanted to do is make sure we don't burden customers with running like this reliable service that has stayed at scale.

And so we take that onus on and it's a, it's a responsibility. We take really seriously our, um, our, our company values are reliable mm-hmm. And developers, developers, developers.

And so anytime we have to think about making any trade offs, you know, reliability is the number one thing. We'll go slow on features, but reliability we never compromise on. Got it.

Um, so we're here at Platform Cut. Yes. This is the, I think the third year for platform cut.

It's obvious why temporal I think would say, Hey, platform engineering's probably a good vertical, right? Because it's a good target audience, but what, what, you know, at some point rubber meets the road and reality meets. Mm-hmm.

Conjecture, conjecture. What, what has been your experience for the platform, Khan? Yeah, it's my first time, and honestly it's great.

First and foremost, I think just the sheer number of people, the community, the, the questions coming up, the sharing of lessons learned, all of those pieces. You know, I I'm so glad that I understand this is the first time in New York City. I'm so glad that the conferences come here and, uh, you know, there is sort of a venue to actually sort of engage and talk about a variety of topics.

Mm-hmm. And also it's clear that, you know, platform engineering is sort of continuing to be challenged. Like one of the things I've seen in my career is like the bar just keeps getting raised on platform engineering, and I think it's high time that there was sort of a, a conference and a venue sort of dedicated to how that bar is shifting at the moment as well.

Right. You know what I find interesting? I'm of an age, right?

Like you said, you've been in this longer than you can admit. You, you talk to a lot of the people here and they'll tell you, well, platform engineering's new mm-hmm. Relatively new thing, right?

Yes. DevOps wasn't scaling and cloud native and, and we needed platforms. Right?

Well, you and I both know, we, we've had platforms, right? For a very, very long time. Right.

Where does old meat, new, new meat olds here, right? Is is, I mean, mean, look, I, I remember being a young person, younger person, and some older people would tell me, oh, there's nothing new under the sun. Right?

There are things change, right? There is new, there is innovation, but how much of what you see and hear at the platform, engineering a platform com, right? And what you do every day at temporal represents sort of really new, or is it, let's call it evolution.

Yeah. Built on what you've been doing for a while now. Yeah, That's a great question.

Um, you know, fun fact, when I was at Yahoo back in 2013, the name of the organization I was part of was called Platforms Really. Right? And this organization, my boss reported directly into Marissa, who was the CEO at that time, right?

So even back in 2013, this concept of platforms and platforms being kind of at the same hierarchy as other functional organizations existed, right? Mm-hmm. Now, I think for me personally, I, um, I tend to look about at like what are the problems and how we're solving them.

I tend not to get caught up too much on like the, the terminology. So I'm not an idealist in that sense. Right?

Right. And I think in terms of the, where does all meet new, um, I feel like especially at this moment in time where we are seeing like another technology shift happening and, you know, how does for instance, AI help with platform engineering? And it's, but the old part here is, it's the same pattern.

It's not going to mean platform engineering isn't relevant. No. It just means that the work we do is at a level that is higher.

Mm-hmm. You know, similar to, for instance, looking at Terraform, it was, you went from shell scripts to EYs to AS code, right? And, and that was, you were solving similar problems maybe at different scale, but the technology made your life easier.

Yeah. And I think that is what, where the, the same old patterns kind of continue. How does platform engineering's life become easier?

How do you look at higher level constructs that abstract out some of the more repetitive things you were doing? To me, that's been like a common thread across all these years. Right?

Agreed. You mentioned AI briefly, right? Well, you just in passing look, a AI's influence and, and, and disruption is profound, right?

Right. And I do mean disruption. It, it's, it's more, uh, real in some areas than other areas, Right?

Eventually, I think it'll be across the board, but today it's up and down, it's spotty in terms of platform engineering and what you're seeing in temporal, how is ai, is it really having an effect yet and it not went? Um, I think I, I think there's a couple parts to that question. Um, one is, um, for temporal itself, you know, we are getting a lot of interest from existing customers and nuance that have like AI initiatives going on.

And the main reason why this is happening is, you know, at the core of it, when you think about building an application on ai, you orchestration is like a very core component. Handling failures, stringing together LLMs and tools, you know, these are like the constructs that a workflow in an orchestration engine solves. And Temporal has had the privilege of solving that for years in like a production battle tested environment.

So one category of things we're seeing is definitely more and more adoption of temporal in building ai, because those four constructs around orchestration are very relevant on the other side of the house. Um, the second part is on platform engineering. There's a couple interesting things.

One is we ourselves are experimenting with using AI for things like run books. Mm-hmm. And, you know, like just, um, there's a certain number of repetitive things that AI would be great at.

Absolutely. Let's do that. And, and that's where we are seeing some, some really interesting sort of patterns emerging and we're seeing value.

And then I think the third part of it is I, uh, have seen some sort of, uh, new companies popping up around using AI for observability. Oh, yeah. Right.

Well, even the established observability companies are using it. Correct. Correct.

And so I think that's an area, uh, that will mature. And I'm sure we'll see more and more value for platform engineering coming out of the, the package tools. But I would encourage folks to actually even just play with like, you know, the, the code gens or the document, like one, one kind of oldest new thing that has not changed is finding information.

Where do platform engineers, like you go to a company and an on-call engineer needs to find something. Uh, there is information sprawl everywhere. That is a use case that ai, That's an amazing use case.

Perfect. Right. Absolutely right.

Hierarchy of the whole thing. Right. You know, I, I had a conversation with someone earlier, it was via Zoom, was it person.

Mm-hmm. But they were talking about AI doing this and AI doing that. And I told them, one of the lessons I've learned is mm-hmm.

Just because you can, doesn't mean you should. Right. And I think that that will be a sign of the e the maturity right.

Of AI is that you, you pick the right use cases where it shines. Exactly. Don't Try to just AI everything.

Right. Because then you're gonna have failures. And then the people will start saying, oh, this doesn't work.

This is a lot of hype. It's not, doesn't live up to it. Well, no, it's, at the end of the day, it's, it's still source a tool and you gotta use the right tool for the right.

For the job. Yes, Exactly. And, and that something that, you know, I think would be, we'd all do well to remember it may not do everything Right.

Doesn't have to find the right Fit. Absolutely. And experiment.

And I'll share another, uh, fun fact is within engineering, we do like a sprint showcase every two weeks, and it's open to everybody in the company. We've added a section in that sprint showcase around like the experiments with ai I did last week. And it's open to anyone in the company to come in and talk about, you know, what LLM they use, what tool they use, what was the problem, you know, just like, absolutely.

We're building those learning circles. And, uh, it's been amazing to hear about, like, just some of the ideas people are coming up with. This has spawned more creative, juicy Right.

Into ideas than Right. Anything I remember since the internet going Right. Commercial.

But again, I think we'll have to see where, where that comes down. Yep. Fifth, we're almost out of time.

Fifth Uhhuh. You know what we didn't mention? Yes.

The temporal website. Yes. io is our website, and as I mentioned earlier, we are an open source project.

Uh, we have a community Slack going. We also have, um, whoops. Are you, don't worry about it.

Okay. Uh, and of course, GitHub, you know, check us out on GitHub, right. com/temporal.

Temporal io. Temporal io. There you go.

Yep. TI want to thank you for sharing and our experience here at Platform Con. Keep up the great work.

We'd love to, you know, we do this a lot. Uhhuh, just in person. Uhhuh, anytime you want to talk, let us know.

It's been a pleasure. Thank you. Thank you.

Ti Soer, SVP Engineering at Temporal. We're live in New York, a platform con Enjoy. Thank you.

Hello, I'm Mike Ard, and welcome to another edition of the Techstrong AI Leadership series. Today we're with Raja and Rahman, who is the CEO for Latent View Analytics. And we're talking about the infrastructure costs that go with AI, because, well, it turns out that it might be substantial and maybe highly underestimated.

Raja, and welcome to the show, Mike. Thank you. Uh, uh, thank you for having me on the show.

It's a pleasure joining you today. I think we're all talking now about how AI is gonna be pervasive, and it probably will show up in every application we use, but it is compute intensive, right? And maybe more so than even traditional analytics applications, and that's gonna cost some money.

So what are we not thinking through properly when it comes to the cost of ai? And, and is this going to maybe be prohibitive in some way? Yeah, this is a concern.

Uh, that's, uh, definitely on the top of minds of, uh, a lot of people. Uh, if you look at the evolution of, uh, analytics initiatives for the last few years, you will see that a lot of the low hanging fruit has been taken. Uh, initiatives tended to be run by, uh, owners of, uh, functional areas or business units.

Whereas now we are talking about much larger initiatives spanning the entire enterprise, which means that you need to pull in data from across the enterprise from all the different silos where the data might be reciting. And oftentimes you're also talking about real time streaming data, right? And making instantaneous decisions and deriving insights right from the top of it.

That dramatically increases the volume of data that you need to handle. And with that come the cost of storage as well as computations. So absolutely, at this point in time, uh, people are looking at how do we manage all of this stuff, uh, at a reasonable cost while making sure that, uh, they're getting the benefits from all the AI tooling and all the developments that is happening in this space.

It seems there's also a fierce debate as to whether or not it is less expensive to do AI in the cloud or in an on-premise environment. You see a lot of people talking about the cost of tokens in the cloud is adds up too quickly, and then I wind up finding out that it's less expensive to run in an on-premises environment, but then again, I gotta go buy new infrastructure. So what's your take on what's going on there?

Yeah. And, uh, I see, uh, a little bit of back and forth happening on that. There was a point in time when everybody was, uh, looking at moving, uh, data completely from, uh, uh, on-prem to cloud, if not public cloud, at least a private cloud kind of a model, uh, because, uh, you could provision both the compute and storage right, uh, on an ask needed basis.

Uh, there's been a bit of a swing back on that in recent times because of the, the cost of tokens, right, that you referred to. I think, uh, it's interesting to see that the hyperscalers as well as companies like Snowflake and Databricks are coming with better models in terms of how the data can be stored and handled, uh, drastically reducing, right? Uh, the, the extent to which you might, uh, need to incur costs on storage as well as compute Databricks, for example, in their, uh, recent data and AI summit that happened a couple of, uh, weeks ago, uh, they talked about how they've integrated the capabilities through their recent acquisition nim, uh, into their platform that separates storage from compute, right?

And, and, uh, dramatically cuts down, uh, the time as well as the access time latency, as well as cost of doing all this stuff. So I see that, uh, there's a bit of evolution that is happening and that will now, uh, determine what directions companies take on this matter. Is tokens the wrong thing to be using to maybe set the pricing models for AI usage in the cloud?

Because it seems rather granular and it's, I have to get a token for input and output. And is there just another way of thinking about it? Yeah, I think, uh, tokens, uh, has, uh, come up as an initial model, uh, given the kind of, uh, computing requirements, uh, that, uh, lums, uh, have, uh, in terms of, uh, addressing, uh, analytics requirements as with, uh, any other, uh, technology.

I'm expecting this space to evolve as over a period of time. We have seen that, uh, other technologies. Now I'm, I'm talking about, uh, even, uh, uh, uh, application, uh, programs like, uh, ERPs in the past, uh, when they would've started out, there would've been a certain kind of a pricing model.

And then over a period of time, uh, it evolves to adjust to what is the most appropriate, uh, kind of mechanism to use. Uh, I'm expecting that there'll be a lot more linkage with the use cases and the business impact and the metrics, uh, that one intends to drive as opposed to input parameters, right? Like tokens at this point in time.

Uh, but this is of course gonna be an evolution story, right? And we are allowed to see how it plays out. There will also be a lot of nuances here in the sense that I need to bring the AI model to where the data is and where it's being created and consumed.

So in a lot of instances, I may train the AI model in the cloud, but I need the inference engine to go run in a local data center or some, maybe even out of the network edge. So is that gonna factor into our cost equations? Yeah, I, I, I believe so.

And, uh, partly this is not just, uh, driven by the cost aspect of it. Uh, there is a great deal of concern around, uh, uh, enterprise level security and governance mechanisms and access controls and, uh, maintaining audit trails from a transparency standpoint. I think that will definitely dictate how, uh, data is stored and used, right?

How much of it is being done in the cloud, how much of it is done locally at the enterprise? Also, both enterprises, as well as service providers like us, are looking at how we can build better semantic layers between the data and then the, uh, and the LLMs, uh, uh, that might be, uh, eventually used for answering things. And these semantic layers can bring in a lot more intelligence right into the particular use case, and also cut down the extent of tokens that are, that needed to be passed, uh, and the compute required right at the LLM end itself.

So I'm, I'm guessing that all of this, uh, will evolve over a period of time. Uh, I'm, I'm seeing, uh, the emergence of high quality semantic layers, uh, on different domains and use cases. And these are being built by enterprises as well as service providers.

And this is expected, uh, to continue in the years as well. Of course, the list of potential projects involving AI in every company out there is probably longer than your arm. But will all these cost issues require CIOs and the executive leadership to kind of narrow down the number of projects that they're gonna actually fully fund because these things are so enterprise wide?

Yeah, I'm, uh, expecting that some of the advantages of, uh, uh, gen AI and the agent TK solutions to start, uh, kicking in a bit, uh, on this front in the last three years in particular, uh, because of the macroeconomic scenario and the uncertainty as well, we have seen that a lot of, uh, larger enterprises, fortune 500 companies have held back on big initiatives that will require a significant, significant amount of spend. Uh, I talked earlier about how it is important to harness enterprise wide data, and typically these mean centrally driven projects, uh, where you go ahead and set up the platform and the infrastructure a little ahead of the co expecting the business cases to come later on. Uh, what we have seen in the last three years is a bit more of a tentative approach that, uh, let's first, uh, uh, get a good hold of the business case and then let's get sponsors for each business case.

And that led leads to a bit of an incremental approach. I'm expecting that, uh, the j and agent KA evolution that is happening will help cut down the cost of other initiatives that organizations are running. We are at least, uh, already seeing, for example, uh, the impact of, uh, uh, these tooling and technologies in cutting down, uh, the effort and time required to do diagnostic descriptive analytics work, for example.

And even in building, uh, data pipelines and predictive prescriptive models. So all of that should hopefully feed in, into the, uh, into the budgetary process with savings and productivity gains emerging from the other work. Uh, the expectation is that, that we'll be able to accelerate a little bit more of the, the platform spend on the infrastructure spends Who's taking the lean on AI infrastructure these days.

Because I think initially when I saw these projects, they were led by a data science team that had somebody in there who knew something about infrastructure, and that's how they built one application. And that was kind of as far as they got. But you could argue that if everything's gonna have some AI component to it, I need some centralized approach to managing AI infrastructure at scale.

So is the responsibility for the infrastructure and the inference engines, especially moving back to an IT department and the CIO is kind of stepping in there, or who's taking the lead? Yeah, I, I, I'm starting to see the trend in, uh, multiple organizations. Of course, it's also dependent on the, on the culture and the background and the context, uh, of the companies in terms of how they have approached not just data analytics and ai, but even IT infrastructure and applications in the past.

Uh, and, and some of the organizations are going through this cultural shift as well. But clearly there is a secure trend in terms of the emergence of the chief AI officer or the chief data analytics officer, and they are starting to bring a little bit more, uh, gravitas into the, into the decision making process, right? Uh, like I said earlier, uh, that certain things need to be done just as cost of doing business.

You can, you cannot just wait to do it on an incremental case by case approach, but we need to take a call that this AI infrastructure is absolutely necessary to compete in the emerging scenario, and therefore there is that appetite, uh, and the budget, right, to go out and spend on creating and building that infrastructure. To your point earlier, there are some advances being made, especially in the cloud, and Databricks being an example thereof. Um, it seems very difficult to make an assessment about where the cost vectors are really gonna be in six months.

'cause the pace of innovation seems to be pretty rapid. So what's your best advice to folks who are trying to figure out, you know, an ROI around an AI project? Yeah, I mean, my advice at this time would be that, uh, don't worry too much about, uh, uh, trying to cost this out over a three year or a five year horizon, because there are definitely several aspects of, uh, uh, of the, of the platform, of the infrastructure and the solution that are going to continuously get less expensive as, as we go quarter on quarter.

Even. Uh, in fact, at the Databricks, uh, summit that I attended, uh, Databricks talked about how several things will now be available native within their, uh, platform like Gemini, for example, right? To do AI a work, uh, or integration with Azure at the backend, right?

Uh, Databricks is also pursuing, uh, a fairly defined strategy where they're saying that you don't need to bring all the data into the Databricks platform. You can leave it where it is in the legacy system, and you can just pull it in right as and when request to do the compute and the modeling. So these models are evolving quite fast in some sense.

I will feel, uh, I feel that we are currently at a point where, uh, all of the innovation that is being done, not only by the, the large companies, but also the startups, uh, they're all starting to converge. And that is gonna help, uh, take care of the cost and the, uh, and the cost benefit aspects of it. So therefore, the advice would be that focus more on the use cases that are going to give the biggest bang for the buck, evaluate it more from, uh, the end output, right?

What are we targeting, right? Uh, is it a top line metric? Is it a bottom line metric?

Are there other business process metrics? And what kind of lift can we see or needle movement can we see in that? And then prioritize on the basis of that cost?

I would expect that given the competitive scenario, uh, they will all trend in a direction where no, it's gonna be beneficial. Aren't we? Kind of bouncing between two extremes here, and let me describe those two extremes.

But, you know, when I first started 2, 3, 4 decades ago now, the wisdom was bring the compute to the data, and then we shifted to the cloud, and we wound up moving a lot of data to the compute. Are we kind of trying to find some middle ground between those two extremes? Now, I would say that, uh, it depends on the, on the use case.

I mean, uh, if you're really talking about, uh, uh, real time, uh, analytics on, uh, streaming data and, and many companies are starting to explore what they can do on that front, uh, there, the technology, the current technology and the tooling that's available might mean that no, you need to bring, uh, uh, the compute to where the data is, uh, know if, if dealing with very large volumes of data, uh, the tooling and technology is evolving there as well. Uh, but overall, I feel that, uh, uh, there is a lot more happening today in terms of bringing the data to the compute, uh, because it's also now possible for, uh, identifying just what the changes are to the data, right? In fact, uh, uh, the separation between storage and compute, uh, a lot of the, the technology that goes into that is really about, uh, how do we identify not just that data element, but also the incremental change that is happening and just capturing that essence so that we don't need to move the entire copy right, of the dataset in order to do the compute.

But they were just able to understand the changes that have been made and then just reflect that, uh, into the compute environment. So I'm expecting that, uh, more and more in the future, data will move to the compute rather than the other way around. But today, depending on the current stage of evolution of the technology and the use case, uh, there will be a balance that plays out.

What do you see among your customers who are getting this right? What are they doing that others are not, that you kind of wish everybody else would kind of be more cognizant of? Yeah, I mean, the one thing that, uh, we keep reinforcing, uh, with, with all the clients, uh, the, and prospects that we are having conversations with is, uh, the need to do a lot more experimentation at this time.

Uh, there is a great deal of optionality, right? That's, uh, evolving. Uh, our vision as a company is to help, uh, company, you know, businesses harness a power of data and analytics to thrive and succeed in a, in a digital world, right?

And that can be possible only if there is a really good understanding of how the ecosystem is evolving and how to cut through the optionality. Just like how there was a systems integrator role in the past over the last 20, 30 years, uh, I believe that there is a need for a AI integrator role, given that, uh, there is optionality at all layers of, uh, data analytics and decision making now, right? From the data layer, uh, to reasoning models and beyond.

Uh, so at, at this point in time, uh, the, the, the trends that I see, uh, in companies that are, uh, really leapfrogging and taking the lead, and this is the willingness to do a lot more of that, uh, experimentation. And that is what is helping them come up and do pilots, POCs at speed, take them from pilot grade to production scale, right? And then start implementing within their environment.

That, and then the other thing I would say is that, uh, the environment, uh, that good that companies create, uh, is very important. Uh, some of the organizations that we work with, uh, they have very clearly defined data analytics at the core of the decision making strategy. One of the, one of the large accounts that we work with, for example, uh, they have this, uh, very, very clear, uh, data-driven operating model that they use that, uh, every meeting and every decision that they make has to be supported by data analytics, right?

That is coming from their data, from their enterprise data and, and their data platforms. So driving that culture of change in terms of saying that, let's look at every decision that we make from a data perspective, uh, plus that, uh, uh, culture of experimenting, right, uh, at large scale. I think those are the differentiators that I see.

Last question. Are you seeing organizations kinda look beyond GPUs or are they looking at others classes of processors to run some of these AI workloads? Or is it pretty much, you know, GPUs or the answer?

What was the question? Yeah, I would say that, uh, uh, at this time, uh, there is the, uh, the, the great amount of, uh, excitement, uh, about, uh, GPUs. Uh, I talked about semantic layers emerging, right?

Uh, in response to the earlier, earlier question, I think, I think those semantic layers and some of the intelligence that get built into the tooling will reduce the need for a, uh, for a battery of GPUs. I mean, there will be use cases, uh, which do not require that kind of, uh, computing power. Uh, it also depends on, uh, what kind of decision is being made, right?

And, and whether the data is being processed, real time, batch, and so on. Uh, that combination with the semantic layer could determine the architecture, right? That, uh, that companies use and, and preference for the compute environment, uh, overall, I think even, uh, large provider, you know, large players like a Databricks or a Snowflake, and the, and the hyperscalers will start providing a spectrum of compute environments, uh, so that again, they can manage the cost depending on the use case that is being executed.

Instead of saying that, now everything needs to go to A GPU. All right, folks, Sharon, and here, I would argue in the last two decades or so, we kind of took infrastructure for granted. I think in the age of ai, if you don't get the infrastructure question right first, everything else is gonna go bad.

Secondly, hey, Rajan, thanks for being on the show. Thank you, Mike. Thanks for having me on the program.

Nice chatting with you. And thank you all for watching the latest episode of the Techstrong AI video series. You can find this episode and others on our website.

We invite you to check them all out. Until then, we'll see you next time. Does AI hate books?

Cisco ICE gets chilly remote flaws. Microsoft is shucking kernel access, hammer space gets cloudy, Intel broker gets busted, and we're gonna make fun of the Department of Justice in this episode of the Tech Fuel Day rundown. Happy summer, everyone.

It's July 2nd. If you couldn't tell by the fact that the sun is 20,000 miles closer to the earth right now, and we are very happy to be bringing you this hot edition of the Tech Field Day rundown. Um, I did wanna point out that, yeah, it's hot outside and it is National Wildland Firefighter Day.

So do me a favor and don't let those guys work anymore than they have to. Um, but I also wanted to mention that it's a national Anna Set Day, and if you don't know what that is, I looked it up for you. It's licorice alcohol.

Don't do that. I know Steven loves that, but he's not here right now. Alistair, as my co-host this week, what's your perspective on black licorice flavor alcohol?

You know, I've not had a large amount of it, and it, uh, yeah, no, no. Let's, let's keep it that way. And I think, uh, I think We're in agreement here.

Yeah. Although your part of the planet is, is 20,000 kilometers closer to the, the sun, my part is not. And so a little chilly here.

So, uh, I'm having some cool, cool temperatures that I'm sure you're jealous of. Uh, I at this point, yes, yes, I am. No matter if they're in science degrees or freedom degrees, I would like them to be a little bit cooler, but don't worry, because we've got some great news headed your way that'll warm the cockles of your heart.

And I'm actually gonna start off with something that may not warm that heart up a little bit, because Open AI and anthropic are taking very different approaches to using books to train their AI algorithms. Andro bought millions of books and then proceeded to cut them all up and scan them into the big behemoth, uh, algorithm. Open AI apparently went a slightly different route, and they use technology to download textbooks from lots of torrent sites.

Neither method seems to be what I would consider to be a traditional method of research, but I guess the question is, we've always heard that, uh, generating access, electricity is killing trees, but now it looks like these models are trying to kill the already dead trees. What's going on? Yeah, it was an interesting couple of stories that I saw, and you'll see in the links that there was two different stories on this.

Uh, one of them was the approach that Anthropic took. Now, apparently, it wasn't their first approach. Their first approach was much more similar to the open AI approach.

Uh, and Thropic got one of the, the guys who used to head the, um, the scanning of books for Google for their project, and said, yeah, go scan all the books. Wow. It's a lot of pages.

Uh, of course, the method that, uh, Google used in the past is non-destructive with automated page turning and automated switching from one book to another. Uh, but that's relatively slow. Uh, anthropic, um, needed a, a much faster method.

And so just cut the spines off and deal with individual pages, and don't try and put the, the book back on the library shelf at the end. Uh, it's this purchase of the books is because you're allowed to do whatever you want with a book when it's finished. But my recollection of going to the library when I was young and such things involve going and looking at books rather than going to the library to get internet and power, um, my recollection was you're only allowed to scan a small percentage of, of the, of a book for a particular purpose.

So it'd be interesting to see just how compliant with fair use it is to cut the book in half and, uh, basically use it to build a huge database, which is what these engines are. These, uh, foundation model engines are, there's a lawsuit brought by authors around the approach that OpenAI took, where they said, well, lots of people have already scanned books, and there's, there's lots of electronic copies available out there. Well, let's go just soak down some of those.

And that seems to have been the illegal distribution, because philanthropic weren't paying the original authors anything. Unlike OpenAI who paid them the pittance that an author gets for each single copy of a book that that gets published. Uh, OpenAI went out and just slurped up every electronic copy of book that they could find on every illegal location that these things were stored, so that there'd be no requirement to pay.

Uh, it seems that there isn't sufficient evidence that individual authors can actually take them to court and say, you've, you've stolen my works and, uh, done this illegally. But probably all of this traffic from these torrent sites has been beneficial to the Torrent site owners because that scraping of, um, these torrents will have involved showing some ads, and that's how the Torrent companies make their money is from showing you ads. So yeah, there's been some interesting approaches to gaining massive amounts of information, essentially all of the written works of humanity in all time.

And, um, it's, there just isn't a precedent for trying to amass that information and use it for commercial gain in the past. Yeah, I still like holding a physical book and turning the pages. Um, certainly my two daughters love the smell of a, uh, page turn on a, um, on a real book, Cisco's issued a warning issued warning that's regarding two maximum severity remote code execution vulnerabilities in the identity servicing engine, ie.

Uh, the flaws impact various versions of the ISE and Cisco strongly advises users apply immediate security patches as there's no workarounds. Um, although there are currently no active exploits that probably won't last for long. Uh, there's also a medium security authentication bypass who's always worrying that's been disclosed.

Uh, although there are fixes out, it's just that customers haven't always deployed these fixes. Security vulnerabilities are always gonna turn up response to it. Workarounds, deploying things.

Tom, are we just talking about being good at doing it helps you here. Alright, stop remediate and listen, because Cisco's back with some brand new contention. I, I had to, I I just had to.

If you run a security appliance, you have to know that that thing is under attack constantly. And especially something like the Identity Services engine ice, it is effectively a gateway product that allows you to get to resources, right? 1 x or, you know, some kind of on authentication mechanism for PPNs and things like that.

So you have to know that if attackers want to try to get the keys to the kingdom, they're going to open up the locksmiths box to try to get them. And that's kind of what ICE does. Uh, two maximum security or pretty highly rated vulnerabilities.

Yeah, that's, that's a thing. Uh, Cisco says there's no workaround you need to patch. Now, this should not be news to anybody.

Uh, in fact, you probably should have been looking at all of the, you know, severity notices coming out. Uh, this came out really closely on the heels of Cisco Live. So I imagine this was probably something that Cisco had known about.

They just, they didn't have anybody available to kind of push the patches out, uh, any sooner. But, you know, it's, it's the nature of the beast, right? And you are right, Alistair, following proper procedures for it, patching and security.

Um, you know, remediation is the best way to do this because it, it doesn't matter what the problem is, right? It could be a zero day exploit, it could be a bug that's been in the code for eight years. It could be a certificate problem, it could be a user escalation issue.

If you are doing what you're supposed to do, you're applying patches regularly, you're testing, you're, you have multiple layers of, uh, defense in depth. The ultimate results of what these things could do is mitigated somewhat, but you've got to make sure that you are working with the vendors who provide these things, and you've gotta make sure you're staying on top of current technology. Because one of the things that will doom you is if you buy this thing off the shelf, you stick it in the corner of your network and everything works, and you just leave it there running, never applying patches, never auditing policy rules, you know, never turning off users that are supposed to be turned off.

Um, those kinds of things can come back to bite you. And you, you definitely don't wanna be bitten by this. Um, it, it'll leave a mark.

There's a vulnerability in the a MI mega rack firmware, which is used by baseboard management controllers. And guess what? It's under active exploitation.

The vulnerability was discovered back in March, and it affects servers from multiple vendors, including Fujitsu, Qualcomm, and Supermicro. The mega rack firmware provides the redfish interface for the remote management system, which allows power control of servers and is often used to install or upgrade operating systems unattended and remotely. Wasn't there a story about some state actor trying to put spying hardware on super micro servers in the past?

Or am I hallucinating that? Well, you might've seen it in a scanned book in your AI system, Tom, but definitely there was a previous story about hardware being inserted in supply chains to achieve what's now being done through redfish. Uh, if you haven't used them before, these, uh, baseband, uh, baseboard management controllers or BMC uh, tools are used for remote management.

We use them a lot where we've got fleets of servers. We attach them to a separate network, a management network, but we use them to do things like taking complete control of the server that they're installed in, but doing it from outside of the server operating system. And so bypassing every security tool you've installed on their operating system, um, vulnerability in these network, uh, these, uh, BMCs, uh, they're network attached.

They have their own independent networking, the, uh, US cybersecurity infrastructure security agency, the CISA, uh, has rated this as a 10 outta 10 vulnerability. There's a authentication bypass that allows access into the BMC at fully, uh, elevated full high privilege, uh, levels that would allow full control also export of all credentials that are stored within the BMC. Uh, this absolutely could be used to provide corrupt images of operating system installs to reinstall an operating system to do a system wipe on the entire server.

This could be used for a massive attack to essentially wipe out all of the physical servers in a, uh, in a customer's network. So, not quite so simple as that because these management networks are never supposed to be directly connected to the internet. Once again, that's good practice.

And if you haven't followed a good practice, then you don't have enough layers of defense against this. But if an attacker has managed to compromise one of the systems that's on your management network, um, can then reach through into the BMC management controllers on all of your other servers, this could be a huge attack surface. Uh, and so protection of these, these networks is important.

What we're seeing is that there are updates for some vendors for their bmc. So updating the firm were to remove this vulnerability. But this is a place where customers are notoriously slow at drawing updates.

Whilst operating system updates are fairly easy updates to the base pain management controllers is not something that, that's usually on that regular schedule. But as Tom highlighted in our previous article, you need to be across the security, uh, notifications and, and the updates that are available for all parts of your environment because every part of your environment can be an attack vector. So definitely this is a pretty critical one.

I, um, always loved having remote management service, so I didn't have to drive in the middle of the night to, uh, to connect and, and, uh, press the power button on that hung server at 2:00 AM I've got flashbacks to that, but as a vulnerability, this is a pretty severe one. And you should look at your servers if you've got redfish firmware installed, make sure it is up to date and that your has provided an update that deals with this vulnerability. Microsoft's hinting it a long-term plan to restrict access to the Windows kernel for third party applications.

SMOs aimed at enhancing system stability and security after widespread, uh, incidents like CrowdStrike outage. Uh, while some security, uh, software relies on critical kernel mode interaction, uh, and has code that runs inside the Windows kernel, this carries a huge risk of system, uh, crashes because a crash in the system, uh, kernel takes down the entire machine, whereas a crash in user mode just take down that one process. Microsoft's developing new capabilities that would allow these applications to operate safely in user mode and gain access to the kernel mode insights through an API that Microsoft provides.

Public statements are pretty vague at the moment. This is an idea that I think Microsoft is floating out and hinting at a gradual transition, but I think they might want to do this fast. Cybersecurity experts believe that the company is paving the way for an eventual ban on direct kernel access, aiming to provide the best of both worlds, a scenario where security tools can function effectively without compromising the kernel's integrity.

Tom, do you need to be in the kernel or is an API good enough for you? No, no, I, there's no reason to be in the kernel. Um, in fact, if you rewind to last year's rundown, when we covered the CrowdStrike fiasco, I think I specifically said, get people out of the kernel.

Um, and here's why I know that you don't need to be in the kernel because other operating systems don't allow kernel access like that. Uh, when you look at Linux or the Darwin kernel, which runs os uh, Mac, os, uh, you, you don't get to play in there. Um, uh, okay, maybe dos or older versions of Windows.

Okay, let's be fair. Nobody's running those either, uh, unless you're an airline. But the, the thing is, there's no reason for those systems to need to run in the kernel.

What happened is, is that the only way to get them to run in the oil in past was to inject them into the kernel, and then, then everybody got complacent, right? Like, oh yeah, no problem. It'll always be there until something happened, right?

And, and we know that the CrowdStrike scenario is the gift that keeps on giving for those of us who write and do the news because it's just gonna keep going and going until someone prevents the system from taking everything down. Years ago at Tech Field Day, we actually had a great presentation from Arista Networks, and they were talking about the fact that one of the biggest advantages that they had in their EOS operating system was that they have isolated and compartmentalized that the demons that run the system, uh, not the little pokey hair, uh, tail, guys with the pitchforks. Now these are the operating system demons, and one of the things that they mentioned was the fact that in other companies operating systems that were mono kernels, uh, something as simple as the status LED uh, process crashing could potentially knock the entire system out.

And Ken Duda the CTO and, and Chief Geek when it comes to coding over there was just like, why? Why is that a thing? Why, why should a status indicator demon be able to take the whole system down?

Well, I would flip that back on Microsoft. Why should anything be able to do that? If you isolate the kernel and allow it to run kernel only processes that you have vetted and verified will not crash the system, and then provide access for these other systems to have that access through APIs, through function calls, through external systems that cannot directly impinge that kernel, then you have created a system that is more resilient to failure.

Look at something like a little snitch. The firewall program for Mac os. It used to be that it did run a kernel module and it required ridiculous permissions to get installed and get removed, and then there was a shift in the way that, uh, MAC OS did network stack interfaces, and they had to completely redo it.

Now, granted, one of the things that that means is, is that when you install an update for a little snitch, you don't have to reboot the whole system because you don't have to update the, the kernel module. Now, think about that. Think about all the number of times that you have to reboot a Windows box.

Well, okay, I'm granted, I'm a little old, but like, all the things you have to do to reboot a Windows box because the kernel needs to be updated because in drivers need to be done. Why just take all that out, leave it sitting over here, not in user land necessarily, but maybe in like super user land or whatever, and that will prevent a lot of these problems. It's, it doesn't take a rocket scientist to figure this out, although maybe if Microsoft hired a couple more rocket scientists, they wouldn't be in the mess that they're in right now.

Hammer Space has launched their tier zero data platform in the Oracle Cloud marketplace to help speed up AI and HPC workloads on Oracle Cloud infrastructure. It allows fast direct data access from on-prem systems to OCI GPUs without the needing to copy all that. Data tests show that it delivers two and a half times faster, reads two x faster rights, and 51% lower latency.

And that helps out because it reduces GPU idle time and supports large real-time data use in hybrid cloud setups. The move strengthens Oracle's position as a very strong option for high performance AI and compute needs. And given the amount of focus that's been put on AI workloads as of late, Oracle wants all that sweet sweet AI cash.

So I guess my question for you, Alistair, is what advantage does Hammer space give to people that wanna run AI and Oracle Cloud? Yeah, everybody wants to pile in for that, uh, that AI money and those, certainly everybody is spending a lot of money to get that AI money and Oracle will have been no exception without OCI buying a whole bunch of GPUs. So getting customers to use OCI for those GPUs is absolutely crucial.

The Hammer Space Tier zero platform is really cool capability to leverage the storage in your GPU hosted servers as a high speed scale out kind of storage, uh, architecture where it functions almost like a case of the data that's on premises and allows you to work with that data that's on premises without actually wholesale copying it up before you can actually start using it. And then also getting that benefit of having usually some nice fast NVM SSDs in your new, uh, GPU equipped service. So this stuff is, is very much about making sure that you can feed data in and out of your GPUs users as fast as possible so that they're not sitting idle as that data is being moved any more than is required.

Uh, generally this is one of the big challenges for reducing the cost for AI training. I don't see this as being used as a, a way of doing foundation model building, but what we are seeing is a large number of enterprise organizations are doing fine tuning. They're taking a foundation model that's been built by one of the, the big players where they've spent tens of millions of dollars and then they're taking that general purpose knowledge and, and adding more knowledge about their organization by this fine tuning process.

That fine tuning requires quite a lot of CPU power that requires a lot more CPU power and particular GPU uh, memory than is available on premises often. And particularly it requires about 20 times as much GPU memory as actually using the model to for inference. So training is a pretty intense activity and best run on the cloud where you can use that resource for a brief period of time and hand it back, but of course, you don't want to have the cost of having all of that corporate data sitting out in the cloud replicating it, the delays of getting it out there as well as the, just the cost of storing it long term.

So this is a good solution around optimizing the use of the GPUs and optimizing the use of your on-premises data without having to transfer that data out into the cloud. This is a really good thing. Nice.

That hammer space is in the marketplace on OCI. So you don't have to do a separate integration to get hammer space Tier zero, uh, deployed on your GPU cluster, and you can make it part of the automated deployment process that you use as you build out your, uh, fine tuning AI cluster for maybe four hours once a week to fine tune your model. That's the kind of use case where the, uh, the scale out, um, access to GPUs and not needing to copy large volumes of data back and forth is really valuable.

Intel broker, the notorious hacker who has caused over $25 million in damages has been arrested in France, known outside of cyberspace as KI West. He faces charges of infiltrating over 40 companies and stealing sensitive data, then selling it on breach forms for over $2 million. He faces charges from the US Department of Justice for wire fraud and conspiracy to commit intrusions and the US wants on US soil, and he wants them ex him extradited from France.

Uh, Kai West was, uh, apprehended through tracking of cryptocurrency, the supposedly untrackable, but actually incredibly trackable way of moving money around. And it was a payment made to a Coinbase account that was linked to him. That was the ultimate, uh, way he was found.

Tom, is there any way to hide from the financial trail if you are wrongdoing? Um, not if you wanna get paid. And I think that's the problem that ultimately happened here, is that, uh, our friend, the Intel broker, wanted to get paid.

Uh, you may remember him from such hits as, uh, hacking into Cisco's dev portal and stealing a whole bunch of information. Uh, he also, uh, jumped into Hewlett-Packard Enterprise and got some of that stuff. And, and then probably the one that I think was the most hilarious is that he hacked the, uh, DC Health Link and then offered to sell a whole bunch of PII from the house of representative members to, uh, to the internet.

Um, but ultimately this was not a crusade to improve security like a white hat hacker. Uh, this was not a statement, a political statement or a manifesto. This was somebody who wanted to get paid and that ultimately was his downfall.

Because look, I I, I've, I've watched enough James Bond movies in my life to know something important. Small unmarked bills might be untraceable, but you do have to put those in a bank somewhere eventually if you wanna be able to spend them, uh, especially in the modern world where everything runs off of digital payments as opposed to cash. And this is ultimately what happened.

He, there was a payment that was made to Intel Broker for some service that he provided or some information that he, he purloined and that crypto wallet was tagged to Kai West, or it might have been Kyle Northern 'cause he had a, an alias. And of course, I mean, if you gotta look for Southern and Eastman and all the other members of the Westing game, shout out to everybody who had to read that in middle school. But it was one of those things where no matter what kinds of protections that you put in, in the front of the house to protect your identity from all of the nefarious things you're doing, uh, all the cops have to do is watch the bank and eventually they will find you and, uh, they will get you because that is ultimately their goal, right?

And, and it was a British national that was living in France doing all of this work. So first of all, I don't know how much longer he could have been there because you know, that whole thing with the eu, uh, he may have had to go back at least a little bit. But now that he's in France, which does have an extradition treaty with the United States, uh, I'm, I'm sure they're going to want to bring him back over here, um, here and find a way to make him pay for everything that he did.

If not in restitution, um, you know, they'll seize his Coinbase account or whatever. Then, uh, they'll just, uh, send jail and he can, uh, enjoy hacking on iron bars instead of, uh, you know, big iron computer systems. Well, it's time for us to take a closer look and it's a closer look at a topic that has been near and dear to our hearts.

As the saga has unfolded, the US Department of Justice has announced a settlement in their lawsuit against a proposed HPE acquisition of Juniper Networks. The release says that in order for the deal to move forward, HPE must sell off the instant online of Aruba networking access points. And additionally, Juniper must auction the license for third party use of Juniper's AI ops for the details of the auction are complicated.

Up to two comp companies can win the rights to license through the bidding process, and Juniper must make up to 55 employees available to be hired by one of the winners to work on the product, including incentives for Juniper for them to move companies. The deal is still pending approval by a judge, but most people are treating it as a done deal. Tom, where do we start digging through this pile of things?

Bottle of tequila? 'cause we might be here for a while. This came out, I believe, either late Friday or Saturday.

I got the email from HP on Saturday saying, Hey, we've got a resolution to this thing. And when I looked at it, I was like, that's really weird. I don't quite understand why they're doing that.

So let's talk about the first part first because it's the least consequential part. A HP Aruba networking needs to get rid of instant on access points. Why?

Uh, it was a line of access points, uh, that basically was for a small business or you know, small to medium enterprise where you only manage like four devices, uh, that you know, realistically you just want the access point hardware. You don't need the, the controller software or any of that other stuff. You're not gonna pay a monthly or yearly feed a license in through the cloud, whatever.

Not sure why they didn't get to keep that, but now they're saying, we gotta get rid of it. And HPE said, sure, no problem. The other thing, and this is where most people have the point of contention.

So when the acquisition was announced, almost everybody that I know said, oh, great, HPE is gonna get Juniper and mist and this is gonna be amazing. And then the Department of Justice came out and said, this is gonna create less competition in the market for wireless companies and we can't let that stand. And a lot of people were scratching their heads going, but really it's because there's more than three vendors in the wireless market.

But okay, and then this remedy comes out and you're not buying Juniper AI ops for mist, you're buying a license to use it. Now, as was pointed out by my friend Sam Clemens, when we recorded an episode of the Tech Field Day podcast about this earlier today, there's no set limit on the amount of time that you get to hold this. And there's no set like restriction other than the fact you got a bid on it at auction.

But as my other good friend Jake Snyder pointed out on that call, you are looking at a minimum of 18 to 24 months to integrate whatever they're doing into your product line. If you weren't already working on this, you probably don't have a plan to do it, which means you are literally starting from ground zero. That's why there was a provision in the auction for you to hire up to 55 Juniper employees.

35 of them are technical assets that know how the product works, and 20 of them are sales assets that know how to sell the product. But HP controls the, the, the code, right? Because you're not buying source code, you're buying a license to use the product.

Okay? Why? So when you, and when you hear from Antonio Neri, which you know, he, he came out with a statement on Monday and he said, well, we told you that we weren't buying this for the wireless stuff.

We were buying it for other things. Yeah, they have a data center switching infrastructure, they have a carrier router infrastructure, they have a campus switching infrastructure. There's a little, a lot more overlap in the campus switching and the wireless piece, but Juniper missed AI ops stuff is infiltrating all of the rest of those product lines.

There was an article that came out earlier this year about how they're applying the, the Juniper ai, uh, Juniper missed AI model to things like carrier routing, which okay, great, I'm down with that. So does that mean that they're not gonna develop that anymore? Does that mean that they say that we could do it better in-house?

Does it not integrate with Aruba Central? I mean, I know that the messaging from Discover last week was, you know, there was a lot of talk of AI and all the things that it's gonna do, but we know that a lot of that is in service to things like GreenLake, where why buy the compute when we can rent it to you in perpetuity, um, like a cloud. So I'm still scratching my head on this, and that's not even counting the conspiracy theories that people have been spending about how the CEO of Cisco flew to DC for weeks on end, trying to convince the DOJ to secretly wreck this deal.

And it was really apparent whenever the Gartner Magic Quadrant came out and Juniper Mist was in the upst rightest part of the quadrant, which means that they must be the bestest of all, and Cisco was actually over here in the wrong spot. And this is all some kind of weird conspiracy theory, and that's pretty much what I tuned out. Um, but Al speaking from someone, as someone who is not intimately familiar with the wireless aspects of this, what do you think is the advantage for HPE pretty much giving in to agree to this ridiculous auction in order to get the rest of Juniper as their new networking division?

Well, I think the simple answer is they didn't think that the missed AI ops was the most valuable thing they were buying, that they, they already have some AIOps capabilities. Ever since the nimble acquisition, AIOps has been a thing that they've been building out through the storage side of the business, and there's no reason that couldn't continue to build out across more of the portfolio. So AIOps absolutely is important, but I don't think that that, um, HPE was starting from ground zero on AIOps and trying to to catch up.

On the other hand, lots of the bits of the market that Juni in are not places that are a huge overlap for HPE. And so having somebody who in those wonderful Gartner Magic Quadrant is up and right in, in the enterprise networking seems like a good place to, to acquire if you're not already up and right in that sector. So I think, yeah, this, um, interesting collection of restrictions put on by put on by the DOJ don't necessarily correspond to actually being a problem for HPE or also being that significant.

I think one of the telling parts on this is that although this is a multi-billion dollar, uh, acquisition, the price that the DO DOJ said was the minimum license price was $8 million. We've already spent more than that on coffee for our lawyers. This, this amount is not anything significant.

It doesn't value that missed AI ops very high if you say the minimum price is $8 million. If you say the minimum price was, I know $200 million, then we're starting to, to think about this being a, a significant piece of the value that's being acquired. But you're very clearly stated that the thing that you had a problem with this operations and, and um, and AI for wifi is not actually there for mobile is not that significant.

You haven't valued it very highly in the overall deal. So it does feel like you are straining on a gnat whilst you're swallowing the camel. Although of course it's a particularly tasty camel, particularly for HPE as they get into another nice piece of market.

And I think, um, my friends at Juniper are, are pretty happy and excited about this as well. Tom is, is there some other place where there's Juniper products that are gonna extend and expand what HP has traditionally done in networking? Yeah, I think that the, the goal of this was the data centers to switching infrastructure they needed, they needed this acquisition for that carrier.

Routers is kind of a, I don't know, it's like IBM's mainframe business, right? Is you are the last buggy whip manufacturer. So you are, you are gonna make money on that no matter what.

You don't have to put a hole on money into it to compete. Yeah, you, you have to put a little in there, but the data center switches are where those real value because it allows them to really focus on building out GreenLake stuff. Yeah.

Alright. The Aruba people are gonna point out that yes, they did sell data center switches and I've never seen one deployed that wasn't in an Aruba HP Aruba network. Like, like people did not go to HPE as the primary vendor for data center switching if they weren't already buying HPE kit.

Um, it's kind of like, you know, in the old days when IBM would sell you everything for a blade center, um, even if the switches weren't branded as IBM, they, they got you ones that kind of really only worked with that system just because they didn't wanna sell you ones from Cisco or Nokia or whomever. So I I, I feel like they're telling the truth when they say that this was not about the wireless part of it. And for the record, um, it is $8 million minimum to get a license for Juniper AI ops for mist.

Uh, Juniper paid $400 million to acquire mist. So it's what, 2%? Uh, at best.

So I, I I think, and someone pointed this out to me, I think they already have a bid in mind. Uh, the, the, the clunkiness like it, it's gonna be two, it can't be three, it could be one maybe, but if both, if more than one company bids over $8 million, then it's the two that they select randomly. We, there's a link to an article from CRN that actually explains this process way better than the order does.

But yeah, it's like, it is, you, you know how it is when people will write an RFP so that only one person can win it. That's kind of what this feels like. And, and I've been noodling on this all day yesterday and today who would buy this?

I still don't have a great answer. And maybe the point is, is that nobody's gonna buy it. They're gonna put it out there for, um, auction and that people are gonna bid on it and they're gonna refuse the bid or whatever and they're still gonna keep control over it and everybody's gonna be happy.

'cause Rami Raheem gets to take over the networking division at HPE and he gets to basically finish out his career there. Um, they become the clear number two favorite against Cisco pretty much everywhere, uh, riding that new momentum from their magic square that says that they're the upst and rightest and the bestest. But I think what's gonna end up happening overall is we've already heard rumblings behind the scenes that there's a coming culture clash with some of the divisions inside of the company where, you know, HP kind of has this stayed, um, traditional approach to marketing and things like that.

And, and these are the upstarts coming in, like, no, we're gonna do things this way and we're gonna do that this way. And I've already heard of, well we've already seen a lot of people who have departed on both sides of the fence saying a lot of the HP Aruba networking people say, well, I'm not really gonna have a place over at Juniper when they come in, so I'm gonna move or I'm gonna leave. And likewise, a lot of Juniper people feel like that they were gonna be replicated.

But the biggest thing I think is that I feel like the Department of Justice almost in a way, they jumped out in front of this and said, well, we have to give them a concession because we halted this. Like we, we absolutely have to say that we did something because if we don't, now we're gonna look stupid. And the other thing is, is that they made it as penalty free as possible because if this acquisition doesn't close, I believe it's by October, then I believe it is HPE that owes a billion dollars to Juniper as kind of the, uh, the, we are serious about doing this acquisition out.

So expect the, the, as soon as the judge signs off this, expect it to accelerate so that they don't have to pay that. But if this ends up being kind of a, a window dressing kind of thing and uh, if there's no real bidder for instant on, people are gonna have to go back and look at what motivated the DOJ to do this and it better be a better answer than we look like we had to be doing something. Yeah, if I'm gonna look stupid, I'd rather look stupid with my mouth shut than say something stupid and have people know for sure that I've done something stupid.

And I think, uh, DDOJ may have done better if they just said, yeah, actually things look good. Hey everyone, future Tom here just got done with an analyst call that was being held because the HPE Juniper Networks acquisition closed on July 2nd and uh, got to hear from Antonio Neri and from Rami Raheem who is now the VP and GM of the combined HPE Networking Business unit. And I was actually able to ask the question, who retains the rights to Juniper AIOps for Mist?

And they asked the question and it was answered. And I just wanna make sure that I update our story here. Uh, the license is just a license for use and support.

Uh, it is not divestiture, it is not a transfer of intellectual property and it is only the AI ops portion. It is nothing else. Um, according to Rami and Antonio, the innovation engine for Mist is incredibly strong now and well into the future.

They have the best talent in the industry and they have over 10 years of experience and learning from real world deployments. So it sounds to me like what's really going on is this was just a small piece of a very bigger hole that was required to be, uh, licensed out or at least the option to license it out. Uh, this is not a divestiture.

Uh, it looks like Mist is gonna be staying with Juniper. And there were actually lots of questions around how Mist is gonna be integrated into all of those product lines. Of course, you know, we're just a couple of hours into the closed acquisition and they're not entirely sure yet, but, uh, just wanted to make sure that we updated the story to let you know that uh, it looks like things are sticking in house for a while and we're gonna have to see where that goes.

But we have a week coming ahead and Tom, you have a, uh, a bit of an event running starting a week today, so you're not gonna be with me on the rundown next week. You be doing something else. You're right.

I'm gonna be back in sunny California and I'm gonna be talking about networking. Um, I mean you gonna be talking about networking with HP Aruba Networking at Networking Field 8 38. We've got a great lineup of presenters that are gonna be taking place on the ninth and the 10th.

Head over to tech field day com. You can see all of the amazing, wonderful, uh, schedule and all the fun stuff that we have planned. And then, uh, right after that, you know, we're gonna take the summer months off or in Al's case the winter months because you know it's gonna be a little warm.

Uh, but we're gonna be coming up on Tech Field Day Extra at Share Cleveland on August 19th and 20th. Steven FoST gonna be, uh, basically driving to work 'cause he's right around there in Ohio. And, uh, he's got some exciting stuff set up for Share Cleveland.

Make sure you check tech phil day com for more detail. Speaking of excitement in September, al, what's going on with you? Well, once the temperatures have dropped a little bit more in, uh, in, uh, Northern California, I will return.

I'll be back out for AI Infrastructure Field Day. Uh, the last AI infrastructure Field day was a massive four day event. Uh, I'm hoping that I don't get quite so exhausted on this new one, but it's already filling up quite well.

September 10th, 11th, probably 12th as well. We'll be back with AI Infrastructure Field Day and then of course, two weeks later, it's back to you again, Tom. Yeah, we are gonna be doing the next edition of Security Field Day.

We are thrilled to have some great presenters lined up already, brand new companies that you haven't heard from at Field day before. And, uh, we've got some other ones in the works, but I don't wanna talk about that because it's secret. com and stay tuned for all of the great information that we're gonna be releasing there.

I promise you it's gonna be a lot of great conversation, just like we always have, not only at Tech Field Day, but each and every week here on the Tech Field Day rundown. Remember we publish our episodes on Wednesdays. Uh, you can consume this as a YouTube video if you are subscribed and have notifications turned on so you know when it goes live.

You can also download us in your favorite podcast application of choice to go on your run, your bike, your swim climb, a volcano, whatever it is you do. And then you can also see the rundown streamed on Textron tv. And, uh, if you have the Textron TV app downloaded, you can catch it there as well.

Um, don't forget to check out the other Textron, uh, TV and Futurum group, uh, productions there as well. We've got a lot of content coming out of Discover Cisco Live, um, you know, a lot of other great places that you're gonna want to, uh, learn about all the cool stuff, not just the technical side of things that Al and I bring you, but the executive perspective and, and a lot of other stuff there will, well, I say we al and a co-host will be back next week, uh, to talk about all the cool IT news that happened. Um, and I'll be back the week after that because, you know, I just can't stay away from this place.

But I promise you I won't be drinking any anset because I, I respect my taste buds too much for that. Um, be sure you tune in and if you wanna leave a comment or, uh, let us know what you thought of this episode, please feel free to do so. We love to read those.

Until next week, take care of yourselves, stay cool and we'll see you for the rundown.