OpenAI Goes Agentic, Drone Crackdowns, and AI Inference Surge | TSG Ep. 1021

Hey, everyone. Happy Tuesday. Welcome to Textron Gang.

We are, uh, still experimenting here live. We're doing it at noon Eastern time. We'd love to hear Is this you like it better at this time?

I don't know. I mean, if you're on the West Coast, I'd imagine you like, probably may like it a little better. You're, you're up and about doing work.

You could have this on versus at, uh, seven o'clock in the morning. But, uh, for others, you, maybe it's the middle of your day and it's hard and you're watching this on demand as a result. But in either way, we'd love to hear what you have to say about it.

You know, as I've been saying, it is the most interesting of times to be covering the tech world. Yeah. Every day brings new, new, interesting and fun things to talk about.

I, I was doing the Security Boulevard podcast this morning, and, and after we stopped recording and we were just chatting, I, I reminded the guests that it's one thing for us to sit up here on Textron Gang or Security Boulevard podcast and talk about these items as pundits and, you know, give our, our thoughts, but we can't lose sight of the fact that this is affecting people's lives. It's a, you know, the, the kinds of things we're talking about are potentially changing humanity's role and how we interact with machines and each other, and technology. And, and so, you know, we, it's great to be glib and talk, but this is some weighty stuff we're talking about.

Um, let me introduce you to our gang to talk about these weighty things. We have Chris Blas, Kate Scar, the man in the High Castle, Mike Vard and, and yours truly. Um, Mike, we're gonna open up with, uh, open Claw meets Open ai.

Yeah. Just when you thought that it was gonna settle down, it's like there's a whole other twist to the drama. Peter Steinberger, who is the guy who created Open Claw, is now going to work for Open ai.

Open AI says that they will donate Open Claw to some sort of open source foundation yet to be named. And at the same time, if you go look over in Security Boulevard, there's a whole cottage industry of folks out there building software that purports to be able to help secure some of these instances of open claw, which, uh, last reports has already been downloaded more than a million times. And people are starting to play with this thing and use it everywhere.

So, you know, Chris, is this the natural order of things? Can this thing be secured or is, you know, this just a matter of time before some form of chaos ensues when somebody hacks into these things and starts stealing people's credentials, which apparently last time I heard are just generally wide open. Why not all three?

Right. You know, so yes, this is a natural order of things. This is how it's gonna work out.

You know, I'm frankly loving this stage of it with Malt Book and all this. Hey, look, we built machines that speak. And Alan, to your point, you know, we've been building stuff in my entire career.

We've been building stuff that impacts people's lives. Like I, I've always loved, uh, the fact that I get to work in OT and IET, right? Because, you know, talk to a linesman, talk to somebody, building water systems and so forth.

These systems, these technologies we've been building since we've built technology, you know, for, for all of human history, affect people's lives. You know, we've been able to feel that we're separating them in it because, oh, you're not really seeing it directly, but, but it, it does. So, yeah, to your points, Mike, it's, you know, we have built these systems that are conversational.

You know, they process words, not just numbers. And, you know, we're a little spooked out by the way they behave. And in recent months, you know, they're starting to get social and we don't like the way that works either.

Right? But that is the natural order of things. That is where it's gonna go.

Now I have my, I have my, uh, opinions about the mechanisms used to do these things to date. You know, they're either gonna be centralized and controlled by I don't know who, and no, we will never know, or they'll be unbounded and chaotic. I don't think either of those are the natural order, right?

I think we're the flaws we're seeing in these stages we're going through indicate where, where we're going to. And you know, Kate, as you said in the green room, you know, I've got my opinions, this canonical base, you know, sovereign, build it from the bottom up instead of the top down. But I, I think it's old, fascinating.

You can see where we're going. And if you think there are problems with this, you're probably right. And those probably can't persist.

Yep. So, Mike, I, I put out an article this morning. It's over on digital CXO.

'cause it's really aimed at how digital leadership should respond to these things. It's called, uh, open claw cows and thing milk. Okay.

My grandma taught me, my grandma used to say to me, you don't buy the cow when the milk is free. She wasn't talking about AI agents, but, but my grandma was a smart woman. I miss her a lot.

But the thing about it is this, it's over 2 million downloads, I believe now. It's actually by, by GitHub Stars. It's the fastest growing project ever.

Right? And so, OpenAI did a move, which is not unique. It's a move that we've spoken about on this show.

Now a number of times, don't buy the cow when the milk is free. They hire, they acquihire the founder, the brain, if you will. And in this case, they didn't even have to buy the IP because it was already open.

But what they did, and, and some may say there's a sinister notion to this, is rather than donating it to like the AI foundation, which is part of the Linux Foundation, which lets you play on a bigger stage with a lot of different characters who, so no one company can control it. And we're gonna talk later on about CCFs 10th anniversary. No, they did it.

They're setting up a separate sort of open claw foundation, which they are going to support, meaning financially support. Now, does that financial support buy them undue influence over this and make it sort of a cottage thing for open ai? Or is it really gonna be a foundation open to the entire industry?

And does anyone else find the irony in open AI taking over what used to be called Claude Bott? Mm-hmm. Right.

And anthropic, Claude. 'cause that's what was it. And they're making it their own.

And, and now they've got this snail, just because they did this, sometimes you could kill the goose that lays the eggs by being too heavy handed. Sometimes you can't catch lightning in a bottle twice. The Peter Sternberger, I believe his name, he may not be able to replicate or duplicate what he did with Claude Bach over for Open ai.

He's in a big, this open AI is a big company with a lot of politics involved. It's a little different than just running on your own with some open source project. Um, but nevertheless, this, this project really may, when we look back, you know, with the little hindsight Monday morning quarterbacking, this may be the project that made ag agentic ai real.

Mm-hmm. Because it kind of sucked last year. I think we all agree a lot of these incidents were not very good.

Here's one I think, go ahead, Mike. Not that though, before you get too far, I mean, it's a nice piece of software and I'm sure it's well done, but I don't see if it's open source, what prevents anybody from replicating it. So Google, Microsoft, and everybody else will have their own flavor of it, and they'll put their own guardrails around.

There's no need to because it, it is pretty agnostic. Mm-hmm. Yep.

Yeah. Well, Kate, you continue. Oh, Chris, oh, I think the only thing that I think from a cybersecurity point of view is that we never architected for autonomous decision making.

And I think this will show a big governance gap mm-hmm. Exploiting it. And, you know, at the end of the day, it's gonna come back as, you know, I think is, oh, AI is evil.

It's doing all these things. And in fact, it's only showing our flawed software and what we created. But Kate, one man's gap is another woman's opportunity.

I am not complaining. I, you know, it's just, it's going to, it's the reality is what it's, I mean, the recourse this in this is just, is just wonderful. Right?

So yesterday, on a aligning foundation, working group call, somebody mentioned that Ray Kurzweil had recently said that there's like $8 trillion in open source. It's unaccounted for in GDP, right? And we know this one, you know, anybody that knows open source, it's like, you know, the, the, the work done by people and teams and so forth, the compensation for that not baked into the fact that we are monetizing at the corporate level a lot of this work.

And that's been, you know, to, to your point, Kate, that's been a flaw we've all been familiar with for a long time. And this, uh, this particular case is a classic example of it. You know, you, you said it right, Alan, is there undue influence in this particular nonprofit and so on and so forth.

And the answer, you know, we have to derive, given current situations is you can't tell, right? And I've been part of these on all sides of them all my career. And even when you're doing your best, you know, if I represent a company A, B, C, you know, I have an interest.

You know, how is that factored in and is it 10 or 12 people in a room coming up with protocols, or is it the entire world? And this is a talk I've given multiple times in the last two weeks, because I think where open source, uh, how that gets dealt with is similar to how we come up with protocols, right? They will emerge organically, you know, from use in the world instead of being concentrated in working groups at, you know, at named, whether it's the Linx Foundation or this new foundation, like they're still wonderful, but we've been developing all these things centralized and that doesn't scale.

So Yeah. And centralized doesn't scale. And that's like one of the most important things.

It actually will, um, create a ano a whole other slew of vulnerabilities if we go in that direction. But, but here's, here's an interesting thing. And I said it before, you know, one man's gap is another woman's opportunity.

Nature rapports a vacuum. What we're seeing, and I'm, I'm inundated with it, our companies, you know, come literally coming out of the sand just to keep with that lobster crab thing, claws, um, that are, you know, here to secure your claw bots, put the guardrails in, help you manage the risk. And, and Mike, as you mentioned on Security Boulevard, there's a, an entire cottage industry of, of companies, and I don't know if I'd call 'em proto companies, maybe, of solutions to, to handle, you know, some of the perceived shortcomings in Claude Bot.

You know, there's, you know, there's religious schools to teach you the church of, uh, crusta or whatever, crusta, um, you know, it's opportunity. And it, and it, and the ecosystem is, is gathering ground. And that's where you look at something like a foundation there and say, wow, with an ecosystem, 2 million downloads in, you know, a couple of weeks.

That's something. Kay, let me ask you something though. Do you think that somewhere there was a bunch of cyber criminals sitting around a table and they would get up every morning and they have their cup of coffee and look at each other and go, can you believe what they're doing now?

Absolutely. Absolutely. No doubt.

You know, I think what we're seeing is this, um, is, you know, and we know this, the exploitation of, of some really, because I've talked about this, you know what I mean? We just have some really bad written code, poorly written code, and yes, I, i I think it's like, you know, there's probably like this grandstands of bad actors being like, yes, yes. Go, go, go.

Well, They gotta strike while the iron's hot though, as well. Right? You know, the salt typhoon keeps coming up, right?

Yeah. Everybody not familiar with that. That's basically just a way to say that the Chinese government knows the gap in everything in, in western infrastructure down to the, to the device level, right?

If your device is configured a certain way, but the absolute known best practice is a different way, they know the difference. You don't if you're running these systems, right. And that's an opportunity for now.

And, you know, but you know, the opportunity to fix that is also much greater now. So, you know, Kate, you're exactly right. You know, this is showing our problems in systems and at the same time, giving us opportunities that we didn't have the ability to address these things 5, 10, 20 years ago.

Yeah. Now we can. And, and I actually don't understand why it is, um, that we can't use the same systems to help create, um, to, to stop those vulnerabilities at stake.

Well, No, I, I think that is the point. You know, think of claw. If you put, you know, claw bots on your system, on your platform, think of them as T cells that are being held in reserve to swarm out on any vulnerabilities and, and remediate patch do things.

Yeah. You know, take, take it back to the biology of it, if you will, to a biological kind of system. And, and that's what you're looking at potentially here.

Yeah. And, and, and frankly, I, I think this is where AI really does, um, the potential for, can really go exponentially as well. Just really helping us clean up things that we've needed to clean up.

And I don't know why we haven't, we seem to add, you know, I think it's just us generally as people, you know, from a hoarding situation, let's try to clean up. And as we clean up, you know, you always go back to those. I, I swear, watch this show as you go back.

It's like, it's back to hoarding. I mean, it's just, I don't know, maybe it's just human nature and, you know, Well, there is that, but look, here, here's the fact. Why is this thing so damn popular?

Why is it, why do we keep talking about it? Why are we fascinated with it? I personally love it.

So, but Go ahead. Because it, it makes every one of us smart enough to use Agent AI to do everyday tasks that we got to do anyway. And, and so, but and isn't that the real promise behind ai?

Yeah. Or one of the real promises behind AI to, to make our lives easier and better. Yeah.

It's really, It takes your job away. But that's another story. The other way to look at that though is to say, we're giving everybody superpowers, but no controls.

Yeah, yeah. What could go wrong? Yeah.

What all what we find out next block is what we find out. Stay Tuned and we'll tell you. But we need to jump to our next, our next segment, and we're gonna bring in our friend Sid nag.

Sid joins us today. Good morning, Sid. How are you?

Or maybe good afternoon already. Depends who you are. How are you guys?

Good. You're right in time here for our second segment, the drone debacle down in El Paso. Um, you know, Mike, I, I thought first I thought it might be UFOs, but No, it wasn't.

Yeah. What exactly went on here? Well, if you think about this story, it's been playing out for the better part of two years.

When it first started out with all those drones that were flying around New Jersey that turned out to belong to the government. And now apparently they were down at the southern border testing drones 'cause they think they want to use them to combat, uh, drug cartels. And, but I guess they had a disagreement with the FAA about the fact that they were gonna test these things.

So the FAA said, we're gonna close the airports at El Paso, why you go do that? And then apparently, you know, with the hell in a hand basket from there. But it also gets curious and curious, or apparently New York City has asked the government if we could take over some of those drones because well, you know, they might be flying around in places where they shouldn't be.

And there could be some confusion, no word in response from the federal government about that. But ca it sounds to me like, you know, we're kinda like going forward with drones with no policies and no plans, and well, you know, is this open claw for drones? What could possibly go wrong?

You know? Right. I mean, oh my goodness.

Like, what the heck are we all thinking? I mean, you know, it just made me think when I was, um, you know, young and I, you know, making money, uh, in the summers, um, between school breaks, I think I had more training, learning how to paint ex extra exterior of houses more so than these people are getting. I mean, they, they can't even identify a balloon between a drone and a balloon.

Like, oh my goodness, we're in trouble. You know, danger Will Robinson. Right?

So, yes, I really believe that, um, that right now, you know, it just shows so many problems in this space, and it's like so much that we see. I mean, you give, you give people a tool and it's like, you know, especially a tool like this. And it's, you know, it's like the wild, wild west, you know, everybody wants it.

And, but the problem is, is that this is critical. This isn't about learning on the job, and it's really, really concerning. We've already seen, and I'm not trying to get, you know, you know, political here, but when we talk about, um, custom border protection, you know, already having so many issues that we have seen out in the field, um, you know, now give them these tools and then having this lack of information, it really is becoming a, a really, you know, true nightmare.

And are we comfortable giving lasered arm tech to law enforcement with no controls? I, I just, this is, you know, sort of going from bad to worse, right? Uhhuh, I'm, I'm pretty sure that bad guys are gonna have lasers and start shooting them down too, won't they?

'cause if they're on the other side of the border, who's to say that they can't shoot it down? 'cause it's certainly Mexican airspace. So it's a different rule.

So I don't know. Yeah, what we definitely don't have policies around this. We definitely don't have any sort of governance around this.

And, you know, for such a long time, I mean, drones are, and you know, Chris, we've talked about IO OT before, right? I mean, if this is not, you know, iot on steroids and us not having any sort of visibility into this iot device, it's a problem. And I don't know what kind of wake up call that we need, um, for us to be able to understand that we have a serious issue here.

And I love technology. Well, well go ahead Chris. Like in the last, like, we ended in the last segment, right?

You know, we're at the this point where we have these problems and maybe we have the, the opportunities, you know, using similar tools to fix them. And in my world, there's a couple simple words, you know, coherence, right? Coherence gets very practical and applied these days because we have the tools to, to, and yes, you know, AI and vector stores and so forth.

However, a story like this, you know, and, and anybody who's worked in these spaces know saying in security and cybersecurity, like, look, you have the FAA and CBP, and you have these different organizations, they have data information they work with that is canonical to them, you know, classic definition. Like they could actually act on that information. But they're different sets and they're not necessarily coherent internally.

And in this case, obviously they're not coherent between each other, even though they're part of the greater, you know, a single greater organization. And historically, you know, people do the best. They can even pronounce spreadsheets.

And we have working groups and coordination calls between agencies, but we also just happen to have the tools these days to actually get coherence in policy and, and, and sensor readings and so forth. So a case like this, we should be able to fix this. Yeah.

I mean, we already have flight aware. I mean, why can't we do something like a flight aware with drone aware or something with letters, you know, just like we have on planes, we have numbers and letters to identify. I mean, it shouldn't be that hard.

And then when we have something in our aerospace that is not identified, you know, then it becomes a different story. But we, we need to, Yeah. Timing, you know, getting coherence among multiple systems is a lot about timing, right?

And, and just, you know, anybody, even if you haven't been been involved with these, just imagine it, you work for the FAA, you work for CBP, there's a lot of policies, a lot of human interaction to get, you know, share information across domains and jurisdictions and whatnot. And, and this is a perfect case. By the time you have done that, you get laser shooting at LAR balloons in, in the real world, right?

And that's the kind of forcing function that'll cause the spending and the, the effort necessary to make that not happen. Yeah. To me it's like, I mean, I'm not iot or, or cybersecurity expert, but like, to me it's a governance issue, which is what Kate, you referred to at the beginning of this conversation, right?

I mean, it seems to me like certain federal agencies have jurisdiction over this domain. Like, like Chris mentioned, you know, there's F-A-D-O-D-C-V-P, et cetera. But you know, what prevents, uh, NYPD, for example, to have the authority over in the air space of New York to run these drones.

I know they're pushing for it. I don't know whether they're actually gonna get the permission or not. But you have this conflict in federal agencies and local agencies.

You have issues associated with legality, safety, constitutional issues, civil liberties, a lot of other things I think that need to be thought about as these, this domain sort of gets some level of, uh, what I would say governance, right? So I think the governance structure is missing, uh, in, in many ways. And there's a potential for a lot of abuse in my opinion, right?

So true that things are gonna get more interesting too. Apparently, uh, department of Homeland Security has a contract out asking, uh, SpaceX to create a voice control swarm of drones. So that basically you can sit there and tell the drones as in a, in an organized fashion, fly, I don't know, maybe hundreds of these things, information to accomplish some goal using a voice command.

I mean, what could go wrong with that? How easy is it gonna be to hack into a voice command system? I don't know, Chris.

I think transparency is important too. I think Chris mentioned earlier the drones in New Jersey, right? I mean, I saw, actually saw those drones from a deck, right?

Mm-hmm. So whatever happened to it, we didn't get any final report on what, what those drones are doing, who launched them, and what ended up happening, right? So these are kind of mysteries in many ways.

So, you know, I, I think, I think this is kind of wild, wild west in many ways, right? So, yeah. Well, it comes back to time again, you know, time to transparency.

You know, to your point, Syd, you know, there's so much information we could get, but you're not gonna get, because by the, you know, you could file and you, it's too much effort. And everywhere we look, I mean, this is, this segment is about an instance where well-resourced organizations could not get the time to transparency low enough between each other to not shoot lasers at Mylar balloons. Right.

You know? And that's just an indicator of, of the fragility of this. I, I, I've had the, the, the great honor and pleasure of getting to know Tom Rich who created DHS, right?

And I asked him, you know, one night and said, you're a small government conservative and you created that, right? And again, it's just another indicator it that, that just getting internal organizations to, to communicate across policy domains inside the US federal government. You end up with this, I'll be kind, massive, bloated bureaucracy, you know, that that is as inefficient as it looks, Right?

That basically should be automated because policy domains inside the federal government shouldn't be opaque, at least to each other. So, and here's, and here's one more piece of irony. So the British military is in Ukraine trying to learn best practices for managing and flying drones.

So maybe everybody should just go see, It's ironic. It's the, it's, it's the state of it, right? 'cause here's the thing, what is the Ukraine Russian conflict war showing us that the future belongs to unmanned drones?

Stop putting money into piloted P flights, pilot flights, right? Yeah. Right.

Good point. Now, ever since drones first became available to consumers, there've been no fly areas, no fly areas where supposedly they, they block signal and everything else, and your drone is subject to being shut down, shut down, shut down as well. So that's not new.

What's new is what we're seeing in Ukraine, which is, it's not some guy sitting with a little operator thing, though. There are some of those in Ukraine, but we're talking about autonomous drones. And when you have autonomous drones that are gonna make decisions to evade attack, let's say, or to be stealthy or to carry out a mission which may be nefarious, well, that raises this, that takes it from the little kid flying a kite, right?

To, to a terrorist flying something into a building, right? And, and so that, that's, that's the, that's the risk here, guys. That's the, that's the game we're playing at.

Yeah. We could talk about shooting Mylar balloons with, with lasers. Yeah.

Idiots are gonna be idiots. Stupid is stupid. But what are we going to do when organized groups have autonomous drones?

You know, Iran sells Kamikaze Jones, North Korea. I don't think North Korea does. I think Iran sells kamikaze drones.

They'll sell to anyone who gives 'em money. Yeah. And what happens when we have someone who wants to do something stupid in this country or, or in Europe or in the west, let's call it first world, and they use an autonomous drone to go do it.

We, we have got to have the ability to take these things down. We have to, we have to, we have to develop that if we don't have it. Right?

I, governance and regulation is always a trailing indicator. It's never out in front of the problem. Yeah.

So you don't, you don't think one good man with a gun is gonna make all the difference here? Well, well, how well it's done with stopping shootouts in schools and everything. Yeah.

We need a, we need a counter counter drone agent. Ai. Well, there's a, you can look at it on YouTube.

Um, slaughter bot, slaughter bots, one word. You know, it's 5, 6, 7 years ago. There's this great, you know, produced, uh, um, forward looking bit of, a bit of wake up call, uh, on autonomous drones.

And at the time I said, well, yes, obviously, you know, those capabilities will come along. Uh, but to your point, Alan, you know, the defensive capabilities have to match them, you know, apocalyptic scenarios keep not working out because there's always multiple sides. Right?

And, and, you know, I know we're coming to the end of the segment, but I think there are answers to this, and it's not, I mean, yeah, sometimes you need to be able to take things down, like literal defenses are what they are. But, but this is the same as everything else. You know, if we can have age, uh, killer drones that can run rogue and, and make really good Hollywood movies, then we can't trust them in Fin AI and financial systems either.

And I think the same sorts of answers apply everywhere. It doesn't, uh, get rid of conflict. But these nightmare scenarios, you know, of gray goo, you know, don't happen.

Well, I think Just kind of goes back to responsible in some ways, the discussion on responsible ai, right? I mean, countries can either trade arms at a, at a government level and agreements to do arms trading, or you have, you know, know people running smuggling guns and arms and, and nuclear weapons, you know, under the radar, right? So what do you want to do?

If you have the proper governance structure, then yeah, it'll happen the right way. But if you leave it sort of, you know, want and let people just do their own thing, and to some degree, governments are buying into that sort of policy about not trying to cover in this whole drone business. And I don't know why, right?

I think that's really the crux of the problem. Because unless you have that agreement and that standards on how you're gonna exchange these, how you gonna deploy these drones, how gonna control these? How you, somebody's talked about licensing drones, you know, all of that.

Uh, you're gonna have rogue actors do crazy things with these things. Right? Well, I think to go back to Alan's point that he said, even at the beginning of, of all of this be beginning before block A, is that we're really talking about serious, um, critical issues that are, you know, and we are just able to talk about it, but there's really serious things happening right now that take some real thought because we're talking about lives that the, this is gonna cost some major lives in both of these blocks, um, that people, until they see it happen, it's, you know, and we can't say, oh, we didn't see it coming.

I mean, that's, that's the truth. 'cause we, we are seeing it coming. We just, we need to act on it more so than anything.

So, Well, you know, the end though, that I suppose, you know, to are costing lies and always have, right? And my comment earlier about, you know, OT and IT, in, in the OT space, in operational technology, these systems will cost lives even when they're working properly. Because as a complex world in it, we like to believe that's not the case.

It has always been the case. You know, if you, you wipe out a bunch of people's bank accounts, um, that will shorten lives and maybe end them, right? So the consequences have always been there.

We're being forced to, to address them. And, And I, you know, I remain, you know, firmly optimistic that in the long arc we solve all these things, it's about reducing harm in the near term. Yes, Chris?

Fair enough. Guys, we're over time. We gotta jump to our third.

Uh, it's a little, hopefully a little lighter, a little happier. It's a happy birthday, happy 10th anniversary to the cloud Native computing foundation, CNCF. Mike, what do we have?

Well, There was a small little event in New York to commemorate that, and Jonathan Bryce was there, along with some of the other leadership folks from the CNCF. And it was nice because, you know, they celebrated the last 10 years, but they were short conversation was more about, well, the next 10 years. And from their vantage point, they feel pretty good about where they sit for largely two reasons.

One is it looks like Kubernetes is evolving into a defacto standard for inference engines for ai. We'll see how that plays out. And then they're also making some assumptions that, well, the AI agents that we're gonna build are gonna be calling more backend services than ever.

And most of that will be through APIs calling through to backend cloud native computing foundation projects. And this is gonna result in more consumption and more usage and more dependency on open source software than ever sit. What do you think about all that?

Well, I think, uh, a couple of, uh, thoughts. One is the higher level, higher order level thought is AI inference will drive increased cloud native software consumption, right? So that's pretty clear.

So, you know, I'm, I'm looking at this announcement or this development as AI inference being the primary driver of cloud native adoption. So especially for inference engines deployed on Kubernetes, now it's positioned as a major new workload, right? Which will accelerate, accelerate cloud native consumption in many ways.

But to your earlier point about, uh, the backhand access being more, more pronounced in this, in this model is also important because, you know, I think we're gonna move away from the GUI access to more of a headless services from that perspective. Uh, so where AI just increasingly invoke sort of cloud native backend services with other traditional sort of interfaces that people have been used to. And I think the third issue is the whole issue on platform engineering, right?

So I think you're going to see this evolving software engineering roles around the platform engineering becomes even more central to how teams build and deployed, deploy application that are, uh, cloud native in nature. So I think there's three sort of talking points here. Um, but it's cr clear that AI g agents and gen AI is gonna drive increased consumption of cl Kubernetes based, uh, cloud native services.

So it's a good thing for C-N-C-F-I think overall because they, they can now sort of claim that they've entered the, to the AI world in a big way, right? Uh, that has impact on not just the application of the AI workloads, uh, impact on headless access. And third is the whole platform engineering functionality.

So, Fair enough. You know, here's the thing. First of all, happy birthday CNCF.

What a great job. I think back 10 years ago to when I first, maybe it was a little more than 10 years ago when I first became exposed to the word Kubernetes was down in Austin, Texas at DockerCon, and we were thinking about doing a, uh, a container, uh, website, you know, a new site. And I went down there and I was trying to figure out a good domain name with the word docker in it.

And I realized using Docker was gonna get me suit, but I went down to Austin and in the middle of DockerCon there was, uh, a little truck of, you know, like literally a stand, you know, at the, where they were talking about this thing Kubernetes. 6. And I came back and I told the team, I said, the heck with Docker, there's this other thing uba, something that everyone is talking about, but not to worry, I don't think it's gonna catch on.

It's really hard, you know, this is why I still gotta work for a living. But, um, It only took 10 years though. So, Yeah.

But, but, but here's the real thing. He, the C-N-C-H-C-N-C-F chief, and I forget his first name, Mike Jonathan. John Jonathan, he's not wrong.

Mm-hmm. AI inference will drive inc. Increased clouds native software consumption because more code lifts all boats, right?

It's that simple. However, the real battle, the real battle is, is the cloud native stack is, we know it now, which is Kubernetes and its friends 200 odd, uh, projects in CNCF is the CNCF lineup with Kubernetes at its core, the real stack for AI computing because AI computing is gonna be the dominant computing stack, right? Where today for Greenfield Cloud native computing is right?

Kubernetes based, is there gonna be an AI equivalent? Will Kubernetes take on these AI orchestration chores and that whoever wins that battle will be the winner in the next 10 years? I think we're kind of looking at this a little bit too much.

Like human developers, and I might argue the following, and AI agents in the future aren't really gonna care about whether the open source software came from a consortium or not. They're basically gonna sit there and say, Hey, this LLM got trained on some open source software, and wherever they found it, they found it and they're not gonna care where it came from. And therefore, you know, what is compelling reason to necessarily commit to being part of a foundation when I just need to stand up open source software, expose it to an LLM, and AI agents will find it.

And that will be my route to market and my distribution and all these foundations, not just the C ncf FI just kind of look at 'em while I go, you know, are they gonna be artifacts have yet to your Kate, what do you think? I absolutely would agree, and I saw Chris shaking his head as well. Um, I, I think that, you know, I actually think we're gonna have more of a distributed environment, you know, especially when we're talking about headless.

Um, I, I'm not confident on the cloud model, and I know that probably sounds, you know, but sounds, You know, crazy. But you know, when you look at what Chris's work is doing around, um, around having more on-prem, i, it, this is where I think it's gonna be going. I mean, we might have to like have like a, a short, uh, connection to, to download something.

Uh, but at the end, I think it's going to be more of like a, a tour type of network onion type of network. I think it's gonna have to be in order to, to, to be able to, to handle the speed. I think it's going to be a more of a distributed model instead of cloud-based.

And so I totally agree with you, Mike, And so I disagree, Alright, Because you need that tension in these shows. Uh, but no, no, I dis but I do really do disagree. Let's take it back to the first sex, uh, you know, segment here.

Open AI starting this foundation for open cloth. And I, I had my dots about it because who is going to rule that? Let's remember a time where open source software was not in the foundational model.

It was in what I call the big brother model, where one vendor basically ran the open source project like their own player piano, right? For their own benefit. And no, Microsoft wouldn't contribute to a, a project that Google's involved in.

Google wouldn't contribute to a project that IBM's involved in. IBM certainly wouldn't contribute to a project that Oracle or Sun was involved in. And we all suffered.

And then along came the Linux Foundation and, and the C-N-C-F-A daughter Foundation, one of 40 foundations, I think the Linux Foundation runs. And it took that big brother element out of it. Now, no one company controls the path of software development.

And yes, we could one off everything with AI and say the hell with it, I'm gonna build it myself and my agent go do it for me. But, you know, you look at something like open clock, 2 million stars, four weeks, you look at some of the big open source successes, why are they successful? Because they, they give people the building blocks.

It's like the, the four, five bases, whatever that we build DNA around, right? These are the DNA of tomorrow's AI applications and functionality. And we don't want any one company controlling them, whether it's open AI or Anthropic or whether they have a constitution or the laws of robotics.

And so these sort of foundational multi, you know, player foundations are a way to keep it honest. Yeah. And I can play my spoken ai.

I don't think an AI agent means we gotta go back to Big Brother whatsoever. I think that's an old construct that's also gone by the wayside. I'm saying there's a whole new construct that has nothing to do with big brothers or little brothers has everything to do with open ai and the AI agents and whoever build them can access.

But everyone wants to orchestrate them. I I, I, I agree With Alan because I don't, I think we should f less focus on the word cloud and CNCF, it's more about creating that compute stack compute. Well, it's more than compute, you know, if you think about AI inference, it's about orchestration, it's about observability, it's about networking, it's about security, it's about policy control.

You want all those discrete elements run in silos in sort of proprietary, uh, or, you know, layers or substrates. I think that could cause a lot of chaos. So I, I'm gonna, I, I'm gonna try to agree with everybody except Mike, but, uh, just to destroy my world, Mike, but I should, but you're wrong too, Alan.

So I, I'm, I'm just trying to be funny about it, but This, You know, so observability was the word, you know, that I've been, you know, jumping outta my skin over here about, because you know, I, I try not to be, you know, religious about the on-prem stuff, but, you know, just like in the firewall world, remember this, you know, Alan, back in the nineties when I got to Cisco in 98, you know, the idea was we're gonna get rid of the firewall appliance 'cause this is gonna disappear. And I'm like, no, it's a psychological thing, right? You people want to be able to go put a coffee mug on the thing that's the edge of their, their, their network.

And we sold billions and, and thousands and whatever those bloody things. And I think it's the same sort of way with ai, right? And we're finding that people like their little pod, you know, be physically right here, have all my stuff and my AI running on it, however, right?

You know, the observability of tying into massive open source and cloud stacks. And, and Mike, you know, where, where I particularly wanted to to pick on you was not, you know, getting open source and not carrying No, no, no. That part, getting open source and caring and saying, I'm an agent out here.

I'm, I'm running on my on-prem stuff, I'm invoking stuff up in the cloud, I'm calling open source, and I specifically care exactly how much about which open source as you told me to back home, right? And if you're not telling me exactly what you're doing, I'm not listening to you. And I won't process, I won't use your code, I won't use your cloud stuff.

So I think it's all observability all the way down. And what that means is observability appropriate relative to who you are. And if the whatever layer above you doesn't meet meet that, you just won't see them, you won't interact with them at all.

So if you choose to be, you know, just, you know, just do anything, and I don't care what it is, eh, maybe that works for you. But I think most people will have observability of, you know, that their choices in what they're connecting to. Were actually saying, I Think headless is one step away from using APIs to invoke services that are disposable.

And I'm, and the AI agent's probably gonna look at three different observability platforms and pick the one that does the job best, and the next time pick a different one. 'cause that's the Orchestration that'll be built in. Yeah, I agree with you, Mike.

I do. Yeah. Yeah.

I hate to do so, but I think I do too. So, and those APIs have to be open. I mean, it can't be just proprietary APIs.

So you can, you can't just talk about headless in a vacuum where Agents themselves have to be open. Exactly. If everybody starts, to your point earlier, every vendor starts creating their own headless access points or access mechanisms, it'll be chaotic.

So, Well, I'll tell you what's chaotic, this discussion. We gotta end it, guys. We're outta time.

What a great discussion though. Sid, Chris, Kate, always Thank you so much, Mike. Great, great stuff.

Even if Chris does disagree with you, you, you're still the man in the high castle. They, they can't stop you. Oh, um, we will be back tomorrow with yet more text on gang.

We've got text on tv, we play following this immediately. Um, so stay tuned. There's so much going on right now.

You can't, you, you gotta do everything you can to stay on top of it. You know, it's a great way to stay on top of it. Subscribe to the tech strong newsletters.

A lot of what we talk about here comes right out of our articles and they're in the newsletters. Go check that out. I think you can subscribe on Lincoln now or on any of our sites.

So do check that out. Thank you all for watching. This is Alan Hummel.

We're out.