Techstrong TV August 12, 2025

Hey, everyone. Did your GPT get smarter? You're watching Textron Gang.

Hi everyone. Happy Tuesday. Welcome to the Text on Gang.

We've got, we, you know, we've got a small gang today, but some goodies here, right down to the, right, down to the core. We've got, uh, Steven Foskett and Mike Azar joining me today on the Gang. And that's really all you need.

Uh, Steven. Mike, welcome. Thanks for coming on.

We, it was a busy, it was a bus. It's been a busy week already. It was a busy weekend.

It was a busy Monday. And here we are on Tuesday. And man, it just seems, you know, one thing we could say about present times is it makes for a lot of good stuff to talk about.

There You go. I don't know if it's good stuff. It makes for a lot of stuff to talk about.

Interesting stuff, controversial stuff and everything else. Um, Mike, let's jump right into it. It was a big weekend where, uh, well actually I think it was last Friday maybe, uh, the, the good folks that shot GPT released five on the world, Right?

But the turmoil through the weekend was rather extensive apparently. So on the plus side, Chad, GT five is supposed to be smarter, almost PhD level, also b um, seems to be less chantic. So it may actually not, uh, type as much silly stuff in terms of blowing smoke up your skirt as it were, and will not be constantly going, you know, well, aren't you brilliant?

And here's this awesome stuff I found that's not true, but I don't know, it's not true. And then, um, finally though, the thing where there was a bit of a flop is, as I understand it, they decided that the premium edition of Chat, GPT would let you choose between models, but that the standard edition that most people use would force you to use chat GT five. And a lot of people flipped out about that 'cause um, they like GPT-4 and some of the other models.

And so the apparently open AI is backtrack a little bit on that, and now giving people more choice in their open AI models. But the question I would have, and it seems to me it's like, has open AI kind of, is this mojo starting to slow down? Because I don't know about you guys, but I hear more people using other LLMs and also, um, and the two I hear most often is Claude from philanthropic.

A lot of developers especially like that one and other folks I've talked to, say, you know, it's just fundamentally more accurate. And then other folks are using Gemini simply because it's in their Google Doc and it's the easiest thing to reach out to and connect to, and I don't have to go log into something else. So Alan, I guess, you know, at one point it was like OpenAI chat, GPT was gonna be the dominant, and now they're just one of many.

What's going on in your mind? Well, You know, this is, this is the life of a missionary. And, and I don't mean someone in the jungles, you know, preach or trying to convert people to, to God or something.

But when you, when you, you are the missionary in a given market, right? You're first to market, you are the one who has to have the machete in their hand, you know, slicing through to the, the, the new trail. You're blazing the trail and then someone comes up behind you and just walks a lot easier over that trail that you've already cut.

And so they catch up to you. That being said, look, I'm a chat GPT user and I've been playing with five all weekend. 0 or from four.

But, um, I'll be really honest with you, after using it 10 times, I kind of was looking to go back to four myself. Now I'm a, I am a paid GPT user, but I didn't see the option to go back to fork. I'll check right after we record this today, but I didn't see, 'cause if it's there, I will.

'cause I, I had, well, It wasn't there, It was not there as of last night and, But it is there as of this morning, It is okay, great, because I was, I was a little, I I kind of had it right where I wanted it, you know, and, and then it just slipped outta my fingers. So, uh, now as I say, there is something like, I use a prompt to make my prompt better, uh, is my way of using GPT. And it worked really well in four oh and five.

Oh, it, it's just changing my stuff around too much. And I, I don't know, I, I, I could see where it is smarter, but I kind of like four oh better now. I also dabble in Claude.

I, I, there are things, Claude is definitely, I think, better in including coding. Um, I don't, I, I'll be, honestly, I don't play as much with Gemini other than when I'm stuck in Google and something's growing on. I very rarely, if at all use copilot.

Um, I, you know, I like perplexity for search and for news stories, but I don't, you know, that's as far as that goes. That's kind of where I am. Well, you sound like a plant because what you just said is pretty much what we've heard all weekend, Alan, really, um, uh, from everybody, um, you know, yeah.

0 and, uh, and everybody freaked out because of what you said. Basically, everything I had that I set up that seemed to work doesn't work exactly the same. Uh, which is no surprise because, uh, AI models LLMs are non-deterministic.

So, I mean, half the stuff I had set up didn't work from run to run anyway, you know, I mean, you know, you, you, you run it this time and it gives you a good answer. You run it that time and it just did. Well, what are you thinking?

You know, it, but with five, it's different enough that basically everything you had set up, um, in a workflow doesn't quite quite work, right? So yeah, as of, um, last night, I think Altman and, and OpenAI, uh, if you're a paid subscriber, you can go in the settings and you can say show legacy models, and then you can still use four Oh, uh, and it's the same four. Oh.

So if you were happy with it, I think you can use it. But you know, like you, um, I, I, I, I end up using, I'm a paid open AI user. I use it for summarization a lot.

Um, you know, here's an article, summarize this, here's a transcript from a podcast. What did they talk about? What are the companies they mentioned?

You know, those kind of things. Uh, that's, that's pretty much what I use it for. Uh, and frankly, three was fine for that.

Four did a great job. Um, five, I'm not noticing a big increase. And frankly, all of my workflow stuff, like you mentioned, um, all of my automated workflow stuff, I'm using Gemini for simply because it's paid for as part of my subscription, and it does fine.

And so, you know, with Zapier and, and, and, and things like that, you know, I've got out that all running through Gemini and, um, you know, it's cheaper and, and it's okay. It does the thing, you know, whereas the thing in this case, again is, is, you know, summarize this article, uh, those kind of things, you know, and you know, it's, it's funny, um, you mentioned as well the tone of five. That's what I've been hearing from people.

You know, GPT-4 would say, wow, you've made an amazing discovery. You've learned the secrets of the universe. You might just be the next coming of God.

GT five is like, yeah, whatever, dude, moving on. Um, and, and, and honestly, I think that's probably better for society. Um, maybe not, maybe people don't like it as much though.

Yeah, No, I, it also is good at suggesting, Hey, I did this for you, you know, this could really use that. And if you like that, let me tell you what else I can maybe conjure up for you. And, and so that is, uh, that is an improvement.

You're right, Steven. And, uh, and it, it is, I I do feel much more like I'm talking to a, a peer rather than a worshiper, you know, if we could call it that. Um, Yeah, somebody said it.

Uh, GPT five sounds like an overworked office mate instead of, uh, instead of a sycophant. So, two things I would like to ask you guys about. One is both of you and myself included, are now using multiple versions of these tools.

Whether we seem to all have at least two, maybe three. Is that how it's gonna play out for most people, or are they gonna lock in on one, or is it gonna be this kind of more, frankly, you know, every app I have is gonna have some sort of LLM built into it, and it's just gonna be a feature. And maybe, I don't really care who's LLM I'm using at the moment, unless I happen to know that it's good at a particular thing better than the other ones, such as maybe Claude is for coding Now.

I wish it was that way, but, um, you know, my experience has been that they're not really, um, interchangeable in many ways. In fact, I, I think Gemini does maybe a better job of summarizing than chat GPT. And since summarizing is what I want, then I'm using Gemini.

But, you know, I mean, if it, I just wish that they were more predictable and more deterministic. You know, I mean, how sometimes, you know, you throw something at one of these LLMs and it comes back with something just completely off the wall and you're like, do you lose your mind here? Or something?

You know, how did this happen? Well, there's no mind. Um, I guess the statistics just went down the wrong rabbit hole.

Uh, but I don't find them interchangeable quite yet. No, I, I think, you know, I, I, I would, I would analogize this to multi-cloud. Why do people use multiple cloud?

You know, meaning public hyperscaler. You know, why do, why does, why wouldn't you just put everything on AWS or Google or Microsoft? Why do you have a little here, a little there, some here and mostly there?

Well, because each of the clouds, they all have their strengths and weaknesses. They may not be super strengths and super weaknesses, but there are certain things that each of them do a little better than the competition. I think we're gonna see the same thing with ais.

Um, and it really depends, like Steven says, he uses it mostly for summarization, right? Gemini does a great job on summarization. It comes with our Google license pack or whatever they call it.

Um, why not? But if, if for some reason you were using Microsoft and it was, uh, open AI or, uh, copilot, summarizations are relatively baseline type of AI function, you know, I don't know if you'd feel terrible not using Gemini for it, even though it does a little bit of a better job. It's when it does a much better job that you say hook.

I'm doing, you know, coding. Claude does a much better job coding. I'm going to use Claude every time I code.

Although we'll see OpenAI is saying this version is much better at coding. And that will, I don't think we thoroughly, we have not thoroughly tested that whole idea. Steven, one more idea.

Is this an odd, some odd way a victory for the people? Because the people spoke up and open AI and maybe learned a little humility here about the fact that there are people who actually have a voice and an opinion about all this stuff. I would be hesitant to attribute humility to OpenAI.

Um, but, but that being said, uh, I do think that maybe OpenAI is at least moderate release, moderately responsible to, or responsive to users. By the way, I remember a similar thing happening previously with, uh, GPT-3 to four transition. And, and you know, where basically users were like, wait, wait, wait.

I still want to be able to use what works. Can I just please use what works? And, um, I I think that the AI companies are realizing that these models are not interchangeable and that they're gonna need to continue to support it.

Um, yeah, I wouldn't, I wouldn't say, uh, humble, but I would say, uh, at least moderately responsive to user needs. And, and frankly, um, they worked really quickly to, to implement this. That's the other thing I'll say is, you know, they, in, in 48 hours, it went from, we turned it off to it's back on again.

And that's, uh, that's pretty responsive over a weekend. Yeah. Yeah.

They deserve a little bit of a credit for That be better than the new Coke. Mm-hmm. True.

That. And I guess the other thing that people are also dinging them for is the whole rollout of this thing apparently was lacked a certain amount of, uh, shall we say, panache. And, uh, there was, you know, a couple of product managers sitting on a, on a, on a video somewhere, and that was about it.

And everybody was kinda like, well, that was pretty ho hum kind of thing. Or, and, and I can't help but wonder in the future, are we gonna have these kind of product rollouts anymore, or is it just gonna be like, new capabilities will be slip streamed in and maybe we won't be putting numerals behind things. And that's become a little passe, Continuous improvement.

DevOps comes to ai. There you go. Could be, could be.

All right. Let's take a break here. We're gonna come back and, uh, talk about our B block today.

What's going on in the data center space. And, you know, we've got the, on this hand on in this corner. We've got the south, the north, the east, the west, everywhere you're watching Textron Gang Discover Textron Group, the epicenter of tech innovation.

We are your go-to for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us will 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 as Alan alluded, there is this big battle starting to emerge between different states and regions for data centers. And you see the Southwest now with Arizona and Texas now pretty much becoming destinations for data centers, and they're battling against Virginia and even North Carolina now. And of course, Steven, your, your very own Ohio Valley now has a few data center projects floating around through it as well.

And is this gonna become the next big battle? But, and, and I'm gonna put a little caveat on there. 'cause we've seen now the folks in Tucson decided not to help give the permits for the next data center project.

'cause somebody I guess woke up and said, we don't have the water for this. But, um, how do you envision all this kind of playing out? Is this gonna be, you know, just money pays for land, or is, or are more people gonna be aware of what's going on?

And there will be this so-called NIMBY effect? Well, you know, right now, um, it's not NIMBYs, it's NIMBYs, right? I mean that people are saying, yes, yes, in my backyard, please build this data center until, until the data center gets built, and then their electric rates go up and their water rates, uh, that go up and their taps run dry and, uh, and they've got pollution like we're hearing in Memphis with X'S data center.

Um, yeah, it, it's, it's interesting because it seems right now that there's this, um, crazy race, especially with Virginia and Arizona and other co you know, other states to try to attract these data centers. Um, and yet there's so much pushback from people. I mean, here in Ohio, for example, our electric rates are starting to go up, not because, uh, the data centers are using more power than we have, but because it's all about the peak.

And, and sort of that last, that last watt. So there's, apparently, there's a public auction process where energy utilities, auction off sort of that, that that little bit of power that you need when everything's really tapped out. And the price of that has gone up by more than 10 x, uh, relative to what it was before the data center boom.

So in other words, it's a hot August day and everybody's running their air conditioners, and your local power utility needs to buy just a little bit more to make sure that there's no brownout or anything. Well, that price is set at auction, and that auction price has gone up to such an extent that it affects the rest of the bill as well, because they have to basically make sure that they've got enough, uh, money set aside that they can also, they can bid against the data centers for that little bit of extra power. So, so here in Ohio, for example, um, the year, you know, most people's electric bills, there was an article the other day that, ha ha, most people's electric bills have gone up about $26 a year because of this little tiny little corner case of what if we need a little more power?

Well, that's a problem for people who are not making a lot of money, which is why, uh, there's been a little bit of pushback and Columbus and, uh, here in northern Ohio. Um, and yet, uh, we find out that the mystery buyer that just bought the former GM plan in Lordstown is none other than SoftBank, allegedly to build part of the Stargate data center complex. Now, soft SoftBank is calling everything Star Bank Stargate.

Uh, they promised that they were gonna put a half a trillion dollars into building this data center or whatever. Uh, they don't have that money. Um, so instead what they're doing is open AI is calling everything Stargate SoftBank's calling everything Stargate.

It's like, it's like almost as if they're trying to convince someone somewhere that they're actually making the investment they promised. And I don't, I don't know how that works, but somehow somebody thinks they're investing a lot of money. But the point is, um, the an LLC linked to SoftBank just bought this Lordstown, um, factory that had been intended to build electric vehicles after being the home of the Chevy Volt and, and things like that.

They, they were gonna transfer this into a, a Foxconn EV factory. Well, instead it looks like it's gonna be a data center, which is not so great because data centers don't provide the kind of jobs that people in northeast Ohio had hoped they would get from these, uh, these factories. And, and, and that's sort of the problem right there.

So essentially, we're in a situation where, uh, com, you know, the governments are, are competing to attract these data centers. They're trying to bring these things in for investment purposes for jobs. Sure.

But they're not bringing the kind of jobs that people had hoped for. They're making everybody pay more for utilities. They're, they're affecting people's water.

I think there's gonna be a real pushback, especially in places like Virginia, where it's already incredibly built out. Somebody said the other day that, um, almost 15% of the world's data center capacity is in northern Virginia, um, Arizona, where they don't have a lot of water, uh, for this. And they're gonna have to, you know, try to figure out how they're gonna do that.

Texas, same deal. I mean, you know, Texas isn't the wettest place on earth. Um, some days it is, but only occasionally not.

That's not, not West Texas where they're building this. Um, and, and, and so, you know, what are we gonna do? I I think there's gonna be some serious blowback here.

And I think some of these politicians that are cutting big red ribbons with great fanfare to attract these data centers are gonna be eating crow in another couple of years when their constituents come back to them and say, wait a second, why is my electric bill so high? You know, I, I hope, I hope we were around to do that, right? You know, when you, you know, somewhere, someone somewhere just wants to drill baby Jill, full speed ahead.

Damn, the torpedoes. I used to think when we would see these apo apocalyptic movies like Mad Max and stuff, that that wasteland was created by nuclear wars, who knew it might be created by overbuilding data centers and everything that these data centers need to, to supply us with the AI that we're going to need to tell us what to do after we go to the bathroom. Um, I had, I think that there is a certain aspect of this that does feel like the, going back to the 18 hundreds and the railroad boom, people were selling, you know, land lots on the assumption that, uh, the railroad was, it was coming in to buy all that, only to find out that the railroad went to some other town, and then that town became a ghost town, and everybody lost money on somebody's real estate scam.

I feel like a lot of this data center activity has the same vibe to it. It's, you know, Except, let me tell you why it's worse though, Mike, because what we haven't accounted for is what Steven talked about. What is going to be the repercussions?

What's the real cost of that data center in terms of clean air, clean, clean water, plentiful water, resources, quality of life? These data centers don't provide the kinds of jobs that a factory does, and they're increasingly more and more automated, and they increasingly need more and more energy. And, and at a time where we're saying, well, let's not use solar.

Let's not use wind. Come on. Okay, so what are you left with nuclear or, or dirty fossil, right?

Or coal heaven for a bit. You, you keep doing that. You, you know, where, wheres the next love canal coming?

That's what I'm waiting for. Are we gonna have l canals again? Because didn't we learn a lesson, you know, runaway industrialization and, and problems.

I I don't wanna, you know, I grew up in a time where you couldn't see the mountains in Los Angeles because of smog. You know, one of the, one of the few things my generation has done is we, we did make progress against smog and air pollution and water pollution and set aside conservation efforts and stuff like this. Are we just gonna go back on this?

Because under this administration, it's drill, baby drill at all costs. Just make sure we get our share of it. That's not the world.

We are gonna live our leave our children. We need people to stand up and march in the streets against this and say, no, no, no. If it can't be done environmentally, consciously, it shouldn't be done.

And this, this Ohio factor, I think is a really interesting, um, example of that, Alan. 'cause. So Youngstown, where it's Lordstown is just on the outskirts of Youngstown, Ohio, northeast Ohio, near Pennsylvania.

Um, steel built on steel, uh, steel industry went bust. Uh, obviously it was a heavily polluting industry, uh, but it provided tons and tons of jobs. The, the, and the city almost collapsed.

Um, Youngstown is the only one of the only cities ever to reduce its size, literally to, um, bulldoze roads and subdivisions and say, you know what? That we're done with that we're gonna, we're gonna make the city smaller because the population was smaller. Uh, this Lordstown plant was touted as, you know, one of the most advanced automotive factories in, in the, in the world.

Uh, they were building advanced cars there. GM went bust the car factory closed. A lot of jobs were lost.

Foxconn stepped in, you know, and, and Lordstown Motors, one of the greatest Ponzi schemes in automotive history. Google it. Uh, Lordstown Motors came in, you know, they raised a bunch of money.

They got a bunch of tax ab basements. They said, we're gonna build electric pickup trucks in this factory. It's gonna provide a bunch of jobs.

Lordstown, you know, Foxconn stepped in. We're gonna make this a contract manufacturer for EVs. We're gonna build the fisker, uh, cars.

We're gonna build, you know, these Lordstown, uh, pickup trucks that you thought Lordstown was gonna build. And, and of course that didn't work either. Um, and now we're turning it into a data center.

It's gonna have, I don't know, 12 jobs. Isn't that gonna be great? You know, it's, it's just such a symbol of, of what happens as you move from, you know, steel to vehicles, to EVs to contract manufacturing, to data centers, all in one building.

Um, they should rename it the metaphor instead of, you know, the, uh, whatever they're calling it. I agree, agree. So, do you think that these data centers are gonna wind up in more of so-called red states than blue states?

in the middle of a red state. But, um, you know, is that how it's gonna split? 'cause more folks in the blue states are gonna be worried about these, you know, consequential issues versus other folks are gonna be in the drill baby Joe camp, and we just may have a, a fundamental split of where the data centers wind up as a result.

Yep. Yeah. And, and I would remind people that there's no real red states or blue states.

There's just purple states. Um, you know, as somebody who was to live in Massachusetts, I spent the week in Colorado last week. Um, you know, these, these, you know, the most liberal state is only maybe 10% more liberal.

You know, the most conservative state is maybe 10% more conservative. But because of gerrymandering and because of the way politics works, you end up with these radical red governments and blue governments instead of, and, and the people are just like, wait a second. So I'm gonna go out on a limb here and say that the people of Texas and the people of Arizona and the people of, uh, you know, Virginia and Ohio, they're not all that different.

And I think they're all gonna be affected by this. And I think there's, there's gonna be a hue and cry. The difference is that the people in the supposed red states, which are dominated by like Ohio, super majority Republican government, are gonna ram this stuff through without listening to what the people want.

Whereas the people in the supposed blue states, which are dominated by, you know, the Democrats are gonna say, hold on. You know, let's think about the environmental repercussions and they're gonna like lose out on this investment. And, um, and ultimately though, they may have the last laugh, because losing out may be the victory when it comes to, uh, data centers.

What a revolt and predicament we find ourselves in. It's just boggled, I mean, to say anything. Yeah.

You know, you might be talking about slavery even, right? Some things don't change. And to, to your point now, even families are fighting about this, right?

The Dale Earnhardt's widow is at odds with her stepchildren over land in North Carolina that might be used for a data center. I mean, it could just wind up tearing whole towns apart in the debate, right? Yeah, absolutely.

I mean, you know, again, we've seen this here where, um, in my town we were gonna build a solar, uh, generation facility, um, should be non-controversial. We got a bunch of extra land over there. Let's just put up some solar panels.

No problem. It became this big political thing like, oh, I hate solar. Why do you hate solar?

What has it done to you? You know? But, uh, but It becomes political.

I don't get that whole thing, the whole anti-solar wind thing, like, dude, I'm not even going to get into it. Causes cancer, kills birds and all the other b******t. But anyway, let's call a break on this one.

We're actually gonna be continuing the discussion though, 'cause our next block talks about the nuclear option 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. All right, folks, as Alan hinted, we are gonna talk a little bit more about data centers, but this time in the context of nuclear energy, there's efforts by the Department of Energy work bring with its various labs. Uh, the one in, uh, I think it's Tennessee is Oakridge, is that right?

Um, and they are working with the, some of the subcontractors to determine if they can speed up the process for licensing data centers and infrastructure, especially for use in AI and other high performance computing applications. Alan, have we reached a point now where to our previous thing is that nuclear maybe is the only option? Well, it's not the only option.

It's not the only option. However, I believe in a, a multi-pronged future in terms of energy. I think solar, you know, in spite of what you may hear from our government, solar dollar for dollar might be our best bet today.

Um, wind not far behind. I think the problem we have with nuclear is too many people think of nuclear as the old three mile island or Indian River, or, or these kinds of things, right? This ain't 1950s nuclear energy we're talking about.

We're talking about much smaller nuclear plants. Much more efficient, much more safe nuclear capacity. And you wanna know what, when we look at the, our options, there's definitely, I thi and I, look, I, people who know me know this.

I'm, I'm far from a red stater kind of person in general, but when I look at energy, I think nuclear has a place at the table. It's not the only option. But I think today's nuclear and, and when we talk about today's nuclear, it's fission still, right?

Today's nuclear fission, uh, power plants deserve a place in our grids, and we can't have it. That it takes a generation to approve licensing for a new nuclear power plant. We've gotta get these things down.

I'm not saying just drill baby drill indiscriminately, turn up nuclear, but you've gotta get these down to a reasonable time without having to jump through all these hoops. I do think it's a lot safer and a lot more efficient than it ever was. I think we also have nuclear fusion on the horizon, you know, and much like, much like, uh, quantum computing, it's always five years off.

But that being said, yeah, Let me use my Linux desktop To check on that. Exactly. But you know it.

But one day it'll be here. We inch closer to it all the time. But until then, nuclear fusion, look, it's not the Cher Noble or, you know, these kinds of things that we could do these in the, the, basically in a shipping container size is from what I understand.

And, and the other thing I will tell you, if you look at the totality of nuclear power use, especially submarines, aircraft carriers, spaceships and energy of generation, the amount of mishaps we've had are, are, are damn near close to no. I mean, there, there's a handful that you could point to in all of these years, and that was all on the oldest technology. Think about all the progress we've made in the last 30 or 40 years since we brought the last nuclear power plant online here in the us.

So, you know, contrary to popular opinion, I am all for this. I applaud the folks at Oak Ridge. Now, I'm not saying rush headlong in and just approve everything, right?

But we gotta do something to make it easier to bring nuclear power online. Well, the, right now, nuclear power is just stuck in a traffic jam, and you can't build anything. I mean, that's the, I mean, the biggest problem with nuclear is because it's so intensely not even just regulated, just tied up with paper that you just can't do it.

There's no, you know, and I've, I've ranted against these small modular reactors, number one, because, uh, they're not actually cost competitive. But number two, because you just can't build them because of regulations, uh, they don't even exist because, not because the technology doesn't exist, but because we, we won't let the technology exist. And, and, and so, you know what I'm gonna say, I agree with you on Alan on this.

I wanna see the regulations loosened up, and I wanna see us be able to try to work with some new technologies here, some new nuclear technologies. I think it's a good idea. I think these things have dramatically improved.

Uh, they can build these things. I don't know. It looks like they're meltdown proof.

It looks like they're safe. It looks like they're reliable. I think it would be good.

Ultimately, nuclear is still gonna be costing more than solar plus batteries. But I do think it's better to have all your eggs not in one basket. And I think it would be smart for data centers to have a, um, a small nuclear reactor to help power the data center.

I would love to see solar panels on the roof. I would love to see batteries stabilizing the local grid. Um, I'd love to see them use less water.

But, um, the point is, I think that we need to have some varied sources of energy, like you're saying. And so I do applaud the administration and the folks at Oak Ridge in looking at ways of cutting some of these, uh, some of the red tape around these things. Because as I said, right now it's just all tied up in red tape and you can't build anything.

Do you guys think, though, that maybe not enough attention is being paid to alternative approaches to rolling out data centers? And now we'll cite, for example, the Chinese are experimenting with dropping data centers in the ocean just to reduce the level of cooling that might be required. Uh, I don't think it's completely solves the energy issue, but, um, there's things that can be done in terms of writing more efficient software.

Maybe we get better processors. But I feel like, uh, right now we're constantly on this notion that says we need, you know, massive amounts of energy today. And then I wonder, you know, six years from now, we're gonna be looking at all these data centers that have no purpose.

'cause we found some better way to do it. Yeah, That's a good point. I mean, are these gonna be the next shopping malls Dotting America, empty big box stores and so on?

I mean, you know, think, yeah. 'cause think about it like, like what you're saying, you know, you can't even sell a previous generation Nvidia, GPU because it is on a power per work basis, not profitable to run it. Uh, what happens when the latest generation GPUs fall into that?

Because the next generation are so much better. What happens if this whole AI thing, maybe it's the Wizard of Oz and there's nobody behind, you know, there's just a little guy behind the curtain. You know, what if, uh, there the whole AI industry becomes bust?

Uh, are the, is this all just gonna be a boondoggle? It could be, it could be. Let me click my heels three times.

And if I wind up with you guys surrounding me in a bed, like I had a bad dream. You were There too, Mike. You were there too, Mike.

Exactly. Look, you know, building a data center under the water, what could go wrong? There won't hurt the coral or anything, you know?

Mm-hmm. Well, Microsoft had a pretty good experiment with that in the Pacific. I, I, I remember that was a pretty neat experiment.

Yeah. Hey, If, if it does go wrong, I got more coral. Yeah.

So Coral doesn't Vote. The the, the thing is, what we're really seeing, and we saw it in the last block and, and it's coming in here again, is a banging of the heads around what we need to do from a modernization progress, if I can call it point of view, versus environmentally conscious and responsible. And there was a time where those two things were not mutually exclusive, right?

We, you know, I don't wanna sound like Al Gore, the father of the internet here, but the, you know, he used to preach this kind of thing that you could have your progress and your green too. And I, I would like to see us think along those lines. It's not all, you know, full speed ahead.

Damn, the torpedoes and damn the environment and global climate change, that's certainly not right. Right? And, and, and for us here in the us we have just totally abandoned.

I I, I'm not sure if they've outlawed climate change from the federal government's vernacular yet, but they're damn close, right? That's wrong. That's just so wrong on so many levels.

At the same time, we do need more energy. We need cheaper energy. We need abundant energy and clean and water, but clean water, right?

Why can't we work these together again? Why? We damn it, we have ai can't AI solve this, you Know, it, it, well, You know, they keep saying that.

They keep saying that AI is gonna give us all these new drugs and solve all these problems for people. Uh, I would love to see them work on that. You and me, both.

You and me both. So, and, and, and to that end, right? Are we focused on the wrong things?

'cause in my mind, maybe we just need a better grid and we need to figure out how the power distribution system needs to work better, because maybe we do have enough power, it's just we can't get it to where We need, we need done batteries. And, and at the same time, to Steven's point earlier, you know, it seems to me, so we have some artificial pricing mechanisms in place that are driving up costs. Well, we, we made those up outta thin air, so maybe we didn't need a different pricing model.

Yeah. Well, and, and just to put some dollar figures on that, I looked it up. Um, nuclear power right now, uh, the levelized cost of energy for nuclear power is anywhere between 27 and $146 per megawatt as compared to solar plus battery, which is 50 to $131 per megawatt.

So nuclear can be as little as half the cost of solar, or as much as almost four times the cost of solar, depending on what regulation, depending on reg tape, basically, Yeah. Regulations. And, and so it's one of those things where we need to start thinking about it a little bit more flexibly.

We need to be a little bit more open-minded here. And, um, and, and maybe it could be cost competitive and it could give us a little bit of diversity. And, and, and, you know, in terms of the grid, having smaller generating sites located more distributed is, is safer.

It's more reliable. It doesn't require as big of a grid. It's, you know, it's greener.

Um, it would be good. We'll see. We shall see.

All right. Hey, gentlemen, thanks for joining Textron Gang today. Excellent discussion.

Mike, Steven, uh, we'll see you on soon. Thank you for watching Textron Gang today. As usual, we'll have our Textron tv, uh, line up immediately following stay tuned this whole week.

We're gonna be, uh, running first time runs a lot of the video work we did over at Black Hat last week, and some good stuff there on cyber and ai, of course, and everything else. So watch those. But until tomorrow, on behalf of Steven and Mike and myself, have a great day, everyone.

We're outta here. Hey everyone, it's a Shimmel. Welcome back here to Techstrong tv.

In this segment, I want introduce you to Hemanchu. Shukla. Hemanchu is the co-founder and CEO of a company called Lightbeam ai.

Let's welcome him. Hemanchu, welcome to Techstrong tv. Hi, Alan, nice to be here.

Nice to have you on, and thank you for joining us. So, Iman, you, before we jump into Lightbeam and what we want to talk about with co-pilot and the risk associated, I, I wanna spend a little bit of time talking about you. You're the co-founder and CEO over at Lightbeam.

And you know, it's up to you to tell us what, what possessed you to go out and start a company and co-founder a company, and be CEO. Let's hear a little bit about your journey to founding seed, uh, founding lightbeam. Sure.

So, uh, Lightbeam is all about simplifying data security. And, uh, if I look at, uh, uh, data security as a problem, it's a very broad area. And, uh, uh, it's a very complex problem, primarily because of lack of automation.

And this is something that we ran into at my previous company where I was working at Nutanix. I was leading the AIOps team there, and we were collecting, uh, customer data. And this customer data got shared with our customer success team, our product management team.

And after some time, we didn't have any clue about where all the sensitive data was, and the sensitive data was in form of, uh, uh, reports. It was in our data lake. It was spread across the whole organization, and we had no clue about where all it has spread across.

Uh, and especially to differentiate between the customer data and the employee data was extremely hard. And then we couldn't figure out how to protect this data. And that's what inspired us to start something which can automate the day-to-day operations part of managing the sensitive data.

And that's what, uh, got us started, uh, with Lightbeam because we found that, uh, using the new AI capabilities, you can automate, build a system that can automate and simpl simplify the data security operations and reduce the risk of handling the sensitive data. And if you see the way the world has been evolving, uh, this is something which is really, really critical that organized organizations need to be really careful about how they're managing customer data. Because it is not only, uh, reputational risk, but also something which, uh, I would, if I'm sharing my data with the organization, I would want them to handle it really carefully.

And this is something which, uh, Lightbeam is all geared towards. I love it. I love it.

Um, when did you find found Lightbeam, by the way? It was founded in, uh, December of 2020. And at that time, yeah, AI was very much in infancy.

All the, uh, models, uh, which were, there were just transformer models, which are the fundamental basis for all the newer generation AI technologies like Charge GPT, uh, were just getting started. And, uh, uh, as we all can see, all these AI models has transformed the world in the past four to five years. Yeah.

So this Lightbeam was kind of a COVID baby, then you guys came, founded right into COVID interesting time Exactly. To find a, found a company, huh? Yes, yes, yes.

That was an interesting journey, how we got started, the initial stages when we were building the company where everyone was working from their home. But, uh, uh, it was tough, but at the same time, exciting because this problem just, just got compounded because everyone was working, uh, remotely. Yeah.

And then, uh, sharing of the data and responsibly sharing of the data for every organization has become a lot more complex over a period of time. And, um, exciting times. Absolutely.

I wanna talk a little bit about DDoS, right? I've, you know, I've, I've been in security myself for 25, 30 years, just back from Black Hat last week. Um, you know, DDoS has been a scourge for a long time, right?

Uh, friends of mine in Akamai and CloudFlare, right? They built their whole, not their whole business, but a good piece of their business was built around trying to stop DDoS attacks. And, you know, DDoS went from, from, uh, a hobby type of attack to professionals and the amount, I mean, you know, today's DDoS attacks, I don't have to tell you there, they're massive, massive, the amount of peak data and, and stuff that, that can come in there.

So we've, we've, uh, it, it, it, and it, it boggles my mind when you think about, you know, how just what was a relatively simple, uh, attack surface, a relatively simple path has become such a, a base for different kinds of attacks and different kinds of security risk and, and everything else. ai, I assume that's the website. Yes.

Yes. And, and before we jump in, we're gonna talk a little bit about copilot and some of the other things. But before we do, for, for people who maybe want to get find out more about lightbeam Engage, what would be your best advice?

Like, what's the on-ramp to engage with lightbeam? So, uh, if you were to, uh, talk about lightbeam, it would be all about how an organization is managing the data responsibly, how the data enters within the organization, how it gets, uh, disposed, or how it gets shared with the people within the organization. Uh, and if you are sharing it outside the organization with your partners, uh, what data are you sharing, uh, with them?

So do you have any clue around the sensitive data, across the disparate pieces, which you might be having, and then having an automation built on top of it, uh, so that if, uh, the data has been residing with you for past seven years or more, are you able to retire the data? Who's accessing this data? Um, uh, for example, if an HR employee is having access to employee sensitive data, that is all fine, but if, uh, uh, a salesperson is having an, uh, access to employee sensitive data, that's a total no-no.

How do you manage these policies around the organization is what lightbeam is all about. Uh, it is all about organizations helping organizations manage the data in a responsible manner. And that's how I would say that, uh, uh, as we grow, go into the AI world, which is there data is going to become more and more, uh, important because, uh, there would be agentic ai, which would be taking actions based on the data which is being fed to them.

And are you having the right data being fed to the AI agents would be the world going forward. And that's where, um, as the data becomes more and more important, companies are, uh, like lightbeam are the ones which are helping organizations manage their data. Absolutely.

Alright, if it's okay, let's change gears a little bit, and I want to jump into Microsoft Co-pilot locking it down. You know, like much of what we're seeing with ai, the, uh, potential for good, the potential to make our lives better, easier, work better is great. But the dark side is, it also represents potential risk and challenges.

And it's why we can't have nice things on the internet, right? Because there's, there are people out there who look in organizations and countries, nation states and everything else who, who look to exploit every new thing that comes down the pike. And this is AI is no different, and copilot is no different.

Talk to us. You know, I think we're all aware of the great things copilot can do for us and help us, but talk to us sort of the dark side here of copilot. What, what's the downside?

So if you look at copilot, it has given immense power in terms of democratizing the usage of data within the organization, because now you can feed the data to copilot, uh, and then you can ask questions. So whatever you use to take hours and hours in terms of digging the data, figuring out what is happening within the data can be answered within seconds. But at the same time, uh, the other part of it is by democratizing the data, at times you are handing it over to people who shouldn't be having access to this data.

So, for example, uh, in an old, uh, older world, I might be a part of a group where sensitive data is shared, but I would not even know about it because it is lying somewhere in some drive. Deep down, I don't have to worry about that part, meaning I won't even look at it. So someone shared with me an employee salary information, which is highly confidential, or their medical records, which were, uh, collected.

Uh, and I won't even get to know about it because I'm not dig looking for that information there. But now with copilot, because it is, uh, looking through every, uh, no and corner, I can just type in a query and it would give me the results and the, this data might be 10 years old, but it is available or accessible to me. So this accessibility of the data, the shadow data, which was there, reciting in organization, is all now available to me.

And that exposes a, a big, big problem for the whole organization, because now if I, by mistake, I made a folder, which was open, and, uh, uh, copilot learns about it, it, uh, uh, collects it, then it is accessible to everyone within the organization. So this is the big risk which, uh, which is always there for organization. It was just not exposed, uh, in the way Copilot has made that information available.

And that's, that's what, uh, uh, is a big challenge, which is there. So for example, I was, uh, reading through, uh, some article where what happened was a company was laying off people and they had just a file name, which was saying, uh, layoff list, and someone searched for it. Uh, although the file was not accessible, the name of the file was accessible to that individual, and that information leaked within the organization.

So that's a huge problem for every company, which is there. So now how do they manage every bit of information being exposed to copilot, uh, in a responsible manner? So part of what makes Copilot so powerful though, is that it has access to that information.

And so by saying, okay, we're not gonna give copilot access to layoff list xls, or something like that, right? It, it takes away from the functionality I think of, of the product. But again, this is why we can't have nice things, right?

Because, you know, what do, there's a given a take there. How does lightbeam help with that? Uh, so what Lightbeam does is it would continuously ma uh, monitor who has access to whose data.

So if you look at, um, the whole, um, world, which is there today, people have, uh, people are talking about who has access to what data. So there's a small difference between who has access to whose data versus what data. So just to, uh, elaborate a little bit further, let's take the case that you have a nine digit number, which is social security number, which is there now this is the what of the data, which is there now the moment, and on its own, it has got no meaning because if, uh, I give you nine digit number, is it truly sensitive?

I don't think so. But the moment I associate that nine digit number with Alan, it has got a whole different meaning because now whole of your identity is exposed to everyone. So there is, uh, whose data becomes a really important part of it.

So what Lightbeam is doing is monitoring who has access to whose data. So for example, going back to the earlier conversation, if HR has access to employee data, that's all fine, but if, uh, uh, sales has access to HR data, that's a total no, no. So this is what Lightbeam is monitoring within the organization continuously.

So you can just, uh, put it on autopilot wherein you are configuring these broader rules in terms of who has access to whose data. And the moment this policy is violated, lightbeam will automatically go and revoke those accesses which are there. So in this particular case, if, uh, you have given access to everyone with respect to the layoff which was happening, or someone else who shouldn't be having access to that light beam will go and revoke that access and clean up the cache, which copilot, copilot might have, uh, built in within it.

And that's way, that's how it is going and maintaining the security posture within your organization. I love it. Very good.

Um, how's this packaged light beam though? Like, so is it by how many users you have? Is it, I I'm just trying to think logically, how would you package and sell this?

So, so it gets, um, so you are asking about what is our licensing model? Yeah, it's licensing. So, so it is, uh, we charge, uh, organizations based on, uh, two different, uh, categories.

One is to say that one is if you're using Microsoft or Google, it is on a per user basis. Uh, and the, or the other one is based on the terabytes of data that you are managing. So for example, if you are using an S3 bucket, you might be having a petabyte of data.

So the cost would be just based on the petabyte of data, or you could be having in different applications using petabyte of data. So the MO model is very simplistic there. We will just be looking at how much is data under management within lightbeam, essentially how much data like Beam is monitoring for access as well as the content there.

And the other part of it is if you're using Microsoft ecosystem or Google ecosystem, because people are used to paying on a per user basis. Yeah. Per user.

Yeah. Correct. So, you know, Makes me laugh of my, you're throwing around petabytes a day are like, they're nothing, right?

Uhhuh, we never, we used to dream about petites, you know, terabytes, gigabytes, I mean, you know, but now we're talking petabytes, like it's every day, it's just a couple petabytes, right? Um, crazy, crazy for those. Exactly.

Exactly. Like the world has evolving so fast and, uh, organizations have been collecting so much of data. In fact, we are in conversations with companies who have collected close to 87, 89 petabytes of data.

So, oh My God, That's kind of a, uh, and now if you go and ask them to say how much is sensitive data within that 87 petabyte, they have got absolutely no clue. No idea, no idea. And not even tagged as such.

This is the funny thing, which is they're like, um, during the early stages of Google, uh, they would count for like, it is a 40 GBO of data, and it's a huge amount of data on which hold of the web search was working. And Now, Now It's crazy. I, my personal Google Drive, you know, from our Google package is a eight terabytes, I think, or something like that.

It's crazy. It, I, yes, I don't even, but yet I use it and I'll probably fill it up, but that's the world. We find it.

And that's why you need a ing, right? Because you don't know, there's so much there. Even you dont know, Even like you, I go to you and ask you to say that, look, do you know what data are you carrying in your eight terabytes of data?

You have absolutely no clue how many places you might have scanned a document. Your passport information might be there. Do you know how many copies of your passport or driver's license you might have created on your own Google Lives?

Which You might have seen. I know for a fact, yes. It's, it's just crazy.

And then if you imagine like, for your mortgage information, but, uh, this is not the place where Lightbeam plays in, but it is a needed for each and every individual also to manage the data. Uh, I agree with you. This is, well, you know, it, it, I forget what the term is, but it's, uh, the more you have, the more you need, right?

Yeah. And that's what it is. Anyway, Chu, thank you so much for coming on and talking with us today.

ai. Appreciate it. Uh, keep it up.

And, and you know what? Come back and visit us. Keep us some abreast of, uh, developments there.

It sounds like, you know, you're doing this five years now, right? It's, you're on your way. Please, uh, come back and thanks for being on Techstrong tv.

Perfect. It was pleasure talking to you, Alan. And, uh, thanks for having us here.

ai here on Tech Drunk tv. We're gonna take a break. We'll be right back.

Hey everyone. We're here again on the floor of, uh, black Hat. You know, it's funny, it's a very interesting show floor.

You walk in and, you know, it's so loud. There's so many bells and lights and whistles and, but we found a little bit of a quiet area. We actually didn't, truth be told, we kicked a guy out who was doing a demo here from Absolute Software.

We nudged in gently. Alright, Christy Wyatt, absolute Software. Here's my guest on, uh, text Drunk tv.

Truth be told, she nudged him Gently, Gently, gently. We'll say gently. So, we're here with Absolute Software, as I say with Christy Wyatt.

Christy, first of all, thanks for coming on. Text Drunk TV with me today. Secondly, here's our audience.

Tell them the Christie Wyatt story. Well, first of all, thanks for having me. Um, I'm the president and CEO of absolute.

I've been with the company for about seven years, uh, software developer way, way back in the day. I've spent many years in Silicon Valley. So, uh, Palm, for those who remember Palm Violets, uh, Motorola, apple, uh, Javas Soft back in the day I was at Citigroup for a period of time.

Insider threat company called Dex. I had a company called Good Technology. Um, so I've done lots of lots of different things and now we're here with absolute Very cool.

Um, you know, assume our audience doesn't know absolute, how would you just give us the absolute story then? I like to say Absolute is the coolest cybersecurity company nobody's ever heard of. We have a very tiny piece of technology that's embedded in the bios and has been for the last 15 years of almost every PC on the planet, really.

And so what that really gives you is kind of an unbreakable connection to that device once it's activated. And so the way we use that is, is a whole host of different ways, but always with a focus on creating endpoint resilience. That's really a category we've sort of created and have been evangelizing for about seven years now.

So we can use this to track and manage devices from the firmware. We can use this to, uh, monitor the health of your overall security posture, heal applications if they stop working. So make sure that your security apps are always working.

Um, we do something called rehydration, which means if, if the OS or the devices become overcome or non-responsive BSOD or ransomware, uh, again, we wake up before the operating system so we can remediate things that may be happening in the device and kind of put you back together all remotely, all without user intervention. And then They needed this. So many, We have a big zero trust product as well in our SSE product called Secure Access.

So we do a lot of different things. And it's all in the bios. It's not on the, is it in the silicon?

Silicon? So it's usually in the un flushable part of the firmware. Um, we do have products across operating systems on endpoints, but we're in the unwell part of the firmware, mostly on Microsoft products.

Our Mac OS and Chromebook products operate slightly differently 'cause their architecture's a little bit different. Yeah, yeah. You know, it's funny, I was walking around people, you meet a black hat around into my friend Alan Friedman.

I don't know if you know Alan. He's from a, he just left csa. Okay.

But Alan is the father of SBOs software, biller materials. His new thing is called hbos. Yeah.

Hardware bill Materials. Yeah. I would imagine this is something absolute software would be perfect for.

So, so we've been doing this for a long time because we are embedded in the firmware and we have amazing partners like, you know, Dell, Lenovo, hp, Microsoft. I mean, there's 28 different, uh, hardware providers that we've been working very deeply with over, over several decades. Um, some of our customers actually activate their, the, this, uh, capability in the bios at manufacturing, which means they can actually track the device from the time it takes its first breath, virtual breath.

It's very still low, uh, yeah. Uh, all the way to the time it reaches your hands. And it also gives you the ability to see a lot of telemetry about what's going on in the device from within the bios that a lot of other platforms wouldn't be able to see.

It sounds it's an amazing tool, an amazing tool. Now I'm gonna imagine you sell directly to PC and Mac and, you know, Chromebook manufacturers? No, no, no.

This is a, this is a commonly held, uh, so, so our great hardware ecosystem, our great resellers of our products, uh, but we sell direct to enterprise as well. In fact, we have over, you know, 18,000 customers, uh, everything from small business all the way up through global enterprise and federal customers. So there's a lot of different ways you can buy from your MSP, from any practically any reseller, from any PC manufacturer.

Um, and you can activate it on any device you have. So, so you don't have to activate it at the time you purchase the product. You can, you can activate us across all of your existing assets, no matter how old.

'cause we've been doing this a long time, But it's built into every bio, not every Yes. But it's built into all of these bios. Yeah.

And it's, should we, could we use the word dormant until it's turned out? Right? Right.

So we don't, we don't see these devices until somebody has, uh, installed, uh, an absolute activated product and it will activate that capability in the firmware. And then from that point, you know, that device is very aware, self-aware, and, and very aware that it is kind of connected to your enterprise. So you can, what we, when we talk about self-healing, right, we really talk about rooted in the hardware self-healing.

So we're not just trying to save ourselves. Like a lot of anything that's running at the, uh, OS and application level is, is really kind of preserve themself. They really can't heal themself.

If you're dead, you're dead. Right? I think the difference between us is, is we're in the hardware.

So, so to us, self-healing means I can rip out the hard drive, put in a new hard drive, and the very first thing that device will do is it will wake up and say, wait a minute, something's missing, and we'll get stood back up. We extend that concept of self-healing to our entire ecosystem of partners. So whether it's CrowdStrike or Tanium or literally any, uh, security or many enterprise applications, we'll make sure that they're always there and always running.

We actually, uh, have a research report we've been putting out for about six, seven years now, called our resilience index. And in that we actually show across millions of devices the actual resilience score for most applications. While organizations may think, Hey, I've installed encryption, I've in installed my XDR across a hundred percent of my install base, it's probably only active and running on maybe 70, maybe 80% if they're doing a good job.

Um, things happen all the time, right? And it's, it's not just about patching and and vulnerability management. It's, it could be the user tampering, it could be just a a, an upgrade got installed.

There's a whole host of reasons why endpoints go dark, uh, Blue Or blue. And you, you really don't have time to send an alert to a human being and have them come. This is really why we believe that the, the, the last point of resilience has to be on the device.

The device has to be intelligent and self-aware. And, you know, we, we've seen such a, uh, an emphasis on resilience lately, right? We, we can't prevent everything true, no matter how hard we try.

Resilience is the key. And it's funny, this is not necessarily new. It's been there.

Yeah. But it was, it's kinda like Dorothy clicking her heels. It was there the whole time, you know, It was there the whole time.

Well, to be fair, I think the company has been doing this for a long time, but the use cases, uh, historically have really been around visibility and control. So it's ironic that in this age where we're talking about some pretty sophisticated threats, one of the biggest things that people really struggle with is, where's my stuff? Right?

Oh, there's a, a big os refresh coming in front of a lot of us. People don't know where their assets are. They don't know what state they're in.

They dunno how to get to them. I think that, um, that's long time been our focus is making sure you always have that hard connection. You know, where it is.

You can remotely manage it and remediate it. It was really about seven years ago, and it was because I had come from another endpoint agent technology company, and too often something bad would happen and we'd say, oh, let's go check the logs. And, and surprise surprise, that device stopped calling in, uh, a couple of weeks ago and nobody really noticed.

Yes, it threw an alert. Yes, somebody tried to fix it, but people are busy, right? We, we don't have enough people to go fix all of these things.

And so the light bulb just sort of went off that, that you have this capability to, to heal things from within. Now this was pre COVID, pre-work from home, pre VSOD event, pre ai. But, uh, but I think that it's even more relevant today if we think about the speed at which breaches and attacks are going to happen.

Our, our last line of defense is at the edge. It's where the fingers touch the keyboard. I, I agree with you a hundred percent.

You mentioned ai, you mentioned, so in my mind, post COVID things changed. The, the world changed. AI has been harbinger of a huge change.

Yeah. Um, how is that playing into the absolute vision? There's, there's a bunch of different ways, aside from, you know, we'll sort of start with, there was this myth.

We all sort of convinced ourselves for like a decade that any data that was of any value to us all lived in the cloud. And that the endpoint devices were just sort of non-intelligent transactional things. Not To the dump terminals Disposable, right?

If you talk to someone and said, I could restore your endpoint device, they'd go, eh, who cares? All my data's in the cloud. I think that for, first of all, that was never true, right?

People were creating all sorts of unique data and insights on the device itself. Second of all, you know, in the age of AI, where you have a lot of new content being created and context being created, your digital twin, your digital footprint, fingerprint, and the point of compromise is probably on that endpoint. You can't afford for the intelligence to be sitting in the cloud.

You can't wait for your next instruction. You know, the attack is gonna happen in five, seven seconds or less. You, you need to find a way.

And, and I think there's a lot of really great innovation going on across the cybersecurity ecosystem about how to move more of that intelligence into agent tools, down to the endpoint to the edge. We're the thing that makes it stick. Not aside from the way that we use our own data and, and, and are applying AI within our own products.

I think the more AI enabled tools you deploy at the end point, the more critical it is that you have that undeletable connection. Here's the thing, we're gonna get some of it wrong. People are gonna trial, and they're experiment.

They're gonna push out new things, things will go go wrong, and they'll go wrong quickly. And so if you don't have that, that digital heartbeat, that connection to that device, I call this participating in your own rescue. Like when you call me and you say, Hey, you're in the bios and everything just went blue, or everything just went black, or we're just, we have a ransomware.

You know, the, my question is gonna be, did you, did you activate us? Like, did, did you, did you turn on the lifeline? If the beacon's on right, there's probably something we can do.

That's a, that's a great way of saying it. So, you know, it, it's funny. So everyone out here watching this actually already has absolute installed, probably it may not be activated.

True Beacon may not be on. So that begs the question, what can they, other than going through channel partners that you mentioned, yeah. Is there anything they could do to turn the beacon on?

So first of all, uh, there's a whole host of different ways you can come see us here at blackhead. com, right? There's, there's, uh, opportunities there.

Literally any PC manufacturer, most channel partners, there's, there's no shortage of both applications that have the ability to, because there's a variety of different ways to turn it on to activate it. Um, so there's a whole host. There's no lack of ways.

com and you'll, you'll all things will be revealed. Excellent. Last question.

What do you think of Black Hat so far? Uh, I, I mean, aside from, it's chaotic as it always, it's, it's, it's, it's, it's chaotic, but I think, um, things are happening so quickly, right? And I think the top of mind for everybody here is just the rate of change, right?

I think it's, we're all very excited about ai. I think somebody in the, in the CISO summit yesterday said the words fascinating and terrifying in the same sentence. Which, which I thought was profoundly true.

I think that as things unfold, um, there's tremendous opportunity. I also think there's tremendous risk, and we're all sort of painfully aware that I have huge respect for all of the constituents that are participating here. 'cause I think everybody's working their hardest to figure out how we all kind of band together and respond to, you know, uh, digital workers and, you know, tainted data in your lens.

It's Territory. It's it's craziness, you know? Right.

You've been around, I've been around. I've seen the advent of cloud. I've, I, I remember when cell phones became a thing.

I remember when the internet became a thing and we all had these, sometimes they were probably rose colored glasses, visions of how great everything will be, but getting my experience anyway, getting from here to there. Yeah. There's always bumps in that road that we don't anticipate or we didn't see coming.

And there will be here too, as you say. I, I think, I, I think people are cautiously optimistic. How does that sound, I or terrifying Ed?

You know, I don't know. I, I, I think it's inevitable. And so, uh, you know, I like to say that there's not a lot of benefit in having what I call the Muppet debate.

I dunno if you remember the Muppets, the two guys on the balcony, like, it's gonna take all the jobs, it's gonna be fine. I, I think the answer is, it's, it's, it's here and it's happening. It's happening with your employees.

It's happening with your customers. And so we're, you know, there's a tremendous opportunity to kind of participate and sit down and figure out what is the best way to apply this, to accelerate our businesses, um, but also to help mitigate the risk. And so there's a lot to talk about there.

I agree with you. A lesson I've learned in 30 years of security. The market doesn't wait for security.

Security has to catch the market. Right. And I think that's what we're seeing here as well.

Absolutely. Anyway, best of luck. com.

com. Check it out. We're here at Black Hat.

We'll have more. Stay tuned. Alright.

Hey everyone, this is Alan Hummel from Textron tv, and we are here at Black Hat on the show floor. We were lucky enough with our media passes to sneak in early before the floor gets too crazy. And I stopped over here at our friend's booth, uh, uh, the Frogs Jfr to check in on them and what's happening here at Black Hat on the Jfr front.

And I've got my friend Paul Davis, who's field CSO at J Rog. If you've watched Text Drunk tv, Paul's been a frequent guest. Hey, Paul, welcome back to Tech Drug tv.

Thank you. Glad to be here and glad to be back in Vegas, a black hat. So we came to the swamp, or the, the black hat swamp.

Yes. Yes. It is green around here.

Yes. Yes. There's the frog.

Um, so Paul, a lot of my audience out here is saying, gee, we know Jay Frank. They're a DevOps company. Yeah.

Yep. They're a DevSecOps company. Yep.

But aren't they a black hat security company? So yes, they are. Right.

This is one of the interesting things. If you imagine today's world, you can't push anything product, software out into production unless you're dealing with security. And security has many different aspects.

Yes, you've got the CVEs, you've got the vulnerabilities, but you've also got the compliance, the regulations, and also making sure you've got trusted software out there in production. So to do that, that's a security thing. And security leaders, that's what we worry about, you know?

Sure. Because if we, if we can't say yes, it's good. That's what wakes us up at three o'clock in the morning when we get the phone call saying something's happened to our software.

So it's a bit of a nightmare. That one. Absolutely.

Um, you know, Paul, obviously the, the theme in no surprise, the theme of this year's black hat is ai Oh, yeah. Is AI this, there's AI that Yes. Here in a ai.

They're in AI everywhere. In ai. Yep.

Um, I, I heard Caleb seamer talk yesterday in a session from Cloud Security Alliance. Yep. And he called, he talked about AI washing.

AI washing, Okay. Yeah. Just slapping AI on Oh, Yeah, Yeah.

You know, whitewashing everything with ai. Yep. Let's talk a little bit about that.

Yes. About how Jfr views ai, ML ops as well. How this is all coming together in terms of the frogs.

So machine learning, ai, generative ai, these are, it's, as you said, it's almost like you can't have a real product in today's world unless you've got an AI tag somewhere in there. Every brochure, everything happens. The problem is, is that it's accelerated so fast that we're losing control.

And what we've discovered is, is that many of our customers have multitude of AI and ML projects going on, but they haven't got the control. So the first thing is, is that, interestingly enough, when we talk, like the attack surface around AI and machine learning, there's two sides to it. There's the development side, and then there's one that's running in production.

From that perspective, you really wanna have a strong foundation. So we started last year. First of all, we discovered that there are actually attacking the data scientists.

They're attacking the data scientists going after them. So if you download like a model, you'll actually try and compromise your endpoint, your workstation, right? They'll also do all sorts of tricks.

And especially like with MPC, with this new, uh, tech, we've recently just published some research saying, Hey, look, if this actually can execute code and in infect system, so that's a newer attack surface. And we've, we've always worried about, Hey, let's put endpoint protection on the workstation. Let's do security.

But we didn't think about data scientists, data engineers. And this is a whole new world around that. And so really what we are seeing is, is companies saying, we've lost control of ml.

We need to start putting control points in place. We need a central place to do storing the mls. We need to do deep scanning.

Because many organizations, I was re uh, watching a Gartner, uh, presentation yesterday, 80% of companies are now investigating how they're going to scan their ML models for malicious code. That's either stuff they brought in from the outside world or stuff that's been introduced accidentally. You know, Paul, one of the problems though is that themes, things seem to be moving so quickly.

Oh yeah. So fast. Oh, yeah.

And it's, it's full speed ahead. Damn. The torpedoes.

Right? And, and so we look at something like you mentioned MPC. Yeah.

MPC was like invented in January, basically. Yeah. Here we are in August.

Yeah. It's already the defacto standard. Yep.

Everybody's coming out with an MPC server, and no one is stay stepping back and saying, what about the security? Yes. And, and you know, unfortunately, as someone who's been in security for 30 plus years, this is a familiar pattern.

Yes. Right? Yes.

Yep. We're always the afterthought. Yes.

Oh, yeah. Security. And, and so I worry, I worry a lot about that.

You know, Gartner, Gartner saying 80%, I'll be honest with you, these days, I don't know how much I believe Gartner, maybe we should file their fire their head, uh, static, uh, statistics keeper. Yeah. But you really think 80% of companies are, are scanning their ml.

They're not. They want to, they're planning. Oh, they're Planning.

Oh, yeah. Yeah. So the actual reality, we just published the state of the union report a couple of months ago, and I think it's like over 70% of the executives think that we're using AI with scanning developers is just a little bit over 50.

So they disconnect them. Now, what's interesting is, is that some people are realizing, especially re compliance EU with its AI laws, they're the tough cookie. They're the people that have teeth over here.

We are slowly getting there. Well, around the rest of the world, we're starting debt. So the regulatory pressures are starting to push down the executives and guess who they look at the security leaders.

Yeah. Hey, how are you gonna handle this risk? And it's like, what do you mean?

Well, you, you responsible for IT security. I mean, You don't think you're getting more budget? No.

Oh, no, no, no, no. More or less. You know, but No, but the thing is, is that what is interesting is they're now bringing in the CISO for those conversations.

And the, I'm increasingly having conversation with CISOs about how do I get control of this? How do I start streamlining? And the interesting thing is that the lifecycle around building ML kind of aligns with what DevOps, because at some point you take that model and it's got to be integrated.

And then one of the most frustrating things I find is they go and talk to something. It says, well, what about your open source? Well, we don't deal open source, we just have libraries.

Of course You do. Yes. They use open source for cleansing the data.

They use open source for. Well, no, There applications are built on open source. Exactly.

Exactly. So there's almost this denial in education going on where we have to tell them that it's ignorant There. Yeah.

But basically what you've gotta have is now control points to make sure you are actually understanding what's ending up in production. And that's what's really scaring me, is this thing of ML models are just being integrated in and the, it's being integrated in all these applications. Applications you build, applications you buy.

How do you get control of that? That's what's really scaring me nowadays. I, I will tell you, it is scary.

The, there's two things here. It's the velocity Yes. And the volume.

Yes. Two vs. Oh, yeah.

And I, as, as counterintuitive as it may sound, I think the only way security folk can can get their arms wrapped around this Yeah. Get their heads wrapped around it, is to use ML and AI Yeah. Themselves.

Yeah. To combat the issue. Yeah.

So you gotta, in essence, fight fire with fire. Yep. Let's talk about what J rogs doing around that.

So what's interesting, we actually just released a, a beta of an MPC. Really? Yes.

Go through, which works with jfr, but it's really clever. 'cause you can ask it question saying, Hey, what vulnerable package do I exist? Et cetera.

For vulnerability management, since we have ton, we have a thing called catalog. It is like the gold mine of vulnerability data around open source and LLMs. And you can go in there, and I've got some companies who are saying, this is our default path before we do a, when we do a triage, we're gonna look at that.

So we have, first of all, we have brought a company called Quack, and we've turned that into frog ml. Yes. And that is basically how do I get a consistent building path for how I actually build out my ML models?

Whether it's the data, whether it's building out the feature stores, whether the experiments, whether it's actually rolling it out in production. We've made that whole journey. So jfr very much focused on the process of building a secure foundation.

Because if you've got a secure foundation, then when you've having to monitor it in production, it's got less attack surface, it's got less vulnerabilities. And you can get proactive on that. So from a, from the AI perspective, we got catalog, we have frog ml, we have our advanced security, we have the ability to store ML models centrally.

And that means we can get access, we can generate audit logs. Really? Yes.

We can show who is downloading or trying to use that ai. So you're fighting fire with fire? Well, I wouldn't say fire.

We are a car. We are like the, uh, Pepto bmo. So the calming input, terrible analogy, but we, it's really trying to put the fire down so it's manageable.

So it's less painful. You don't get burnt from it. Because as I say, the more visibility we have into the process of how something ends up in production, the better.

Yep. I wanna bring up another subject with you. Yes.

Lot of stuff going on around platform platform engineering. Yes. DevOps and platform engineering working together for the internal developer platforms.

Yes. And, you know, the developer becomes the customer, if you will, of the product. How does that play into all of this ML ops and, and Right.

You know, the, the base jfr offering, actually. So it is all about streamlining. Most executives want to streamline what, to simplify the process without compromising their security.

The problem is, is that at the moment, because everybody has their all favorite tools, they don't, they can't streamline it. Yeah. But we, we can't beat all things to all people.

So the platform play from the perspective of a DevOps platform, from designed all the way into production. We support that. And we have also the ability to actually, uh, use evidence files as gates to prevent something into production.

If it hasn't passed a test. J link is sexy. Yeah.

Right. But from the point of view of the, the, the developer and enabling them, that's a big thing for us. Shift left.

We've been doing shift left and focusing for years. The thing for me is I want to turn developers into Security Warriors. I want to give them the information at their fingertips.

So you say, Hey, this is rarely a problem. It's like, we rarely need to have traceability. We need to have, is this really a real threat to your software?

There's terrible saying. I have, I use it multiple times. One bad function doesn't make a bad package if you're not calling the bad function.

I don't need to worry about it. I don't wanna tell the about, Hey, you've got a problem 'cause you're using a bad package, but I'm not calling the bad function. Why are you bugging the hell out of me?

I just wanna write code. Yeah. And you talk about, interesting enough, the impact of this across the whole organization.

When a developer makes a mistake and it includes a bug or a secret or whatever, that ripple effect goes to the AppSec team. The AppSec team, they might miss it or they might, they might get through. Then it goes through to IT ops production and then SecOps.

So you have all these IT ops, BizOps. So we actually call it every ops, but I've got to grips other than that, every ops, because the impact of making a mistake at the beginning ripples through you double the cost of deploying somebody deploys a bug Double. No, I think it's more than double.

It's exponentially. Yeah. Yeah.

It's just crazy. So if I can fix it earlier, just like the kill chain, the sooner the fixer, the cheaper it is. And as I said, the fact that in the AI world that CISO's teams are being brought in on the design, I still worry about data.

I don't think we've got data governance under control yet. No. But as far as the actual binaries, the open source, that sort of stuff, I think that's sexy.

Excellent. Um, just about a month out from now. Yes.

We will be in Napa at the Lumbo. We'll, oh, yes. Excited for that.

We'll be there actually filming live. Wow. Brilliant.

Yeah. So we'll be there. I'll probably see you there and we'll talk some more.

Oh yeah. I'm actually doing a training calls there. Very cool.

Okay. So they've got me recorded, like as Morpheus from the, um, the Matrix. Oh, yeah, exactly.

You got, right, because I've got a training course about compliance and showing how you can automate. Very cool. Easiest way security to automate.

But yeah, it's a great event. I went there last year for the first time. It's rarely like the old fashioned security communities where you actually get a chance to sit down and talk.

Yeah. He actually, it's not just No, I, I've been going to swamp boats for Years. An amazing, a great Community Event.

Yeah. So you build long lasting connections there. So thank doubt about it.

Yeah. Looking forward to it. So that starts, I wanna say it starts the eighth September.

Yes. No, it's ninth and 10th. Yeah.

Yep. You're testing me now. I'll have, Yeah, no, it'll be there.

The ninth and 10th filming. Yep. Paul, thank you so much for giving us a peek into j Rog here at Black Eye.

Enjoy the rest of Black Eye. Thank you so much. All Righty.

Okay. Paul Davis Field CSO for J Rog here at Black Hat. We're going to continue our coverage on the show floor.

So stay tuned for now, though. That's out. That's it.

We're out. We'll be back. Hey everybody, welcome back to the six five Summit 2025.

And our topic here has been pretty constant. And it's about making AI real inside of enterprises. Daniel, it's been a great event so far.

I mean, we're covering enterprise AI from silicon to SaaS, and pretty much everything in between. Yeah. It's been a great event, pat.

It is very exciting to kick off our fourth day. We know over the last couple of years, AI has been the central focus of the technology industry. And basically now it's coming to every industry.

Every business is focusing on how to unleash AI to make it part of their business. More efficiency, more growth, more productivity. It's gonna be another great day.

Yeah, for sure. And what we've seen is a couple different approaches. I mean, we've seen AI as a bolt-on to what people are doing, kind of splitting between this modernizing and ai.

And then we've seen, uh, AI first where we've seen a lot of people come, come out and say, we are AI first. Where we architecting as much as we possibly can to do that. And that's what we wanna dive in here, uh, with seven A from OpenText, seven A, welcome to the summit.

Well, thank you for having me. Very exciting to be here. Yeah.

Seven A. It's so good to have you opening up day four for us. It's been an amazing event.

Such an esteemed group of speakers, and you, I expect to bring it for our audience today. So let's start off talking about a little bit of the, the theme that you heard from both Pat and I in the buildup here. You know, we talked about how this kind of AI conversation is evolving rapidly.

We've gone from this experimentation phase to now it's really implementation. You know, where do you, you know, OpenText has had a, a, a lot of product it's been launching in this space services to enable and make this real for companies. Um, but as you've been part of this leading product for all of OpenText, like where do you think the enterprise market is right now in that journey?

Yeah, well thanks for having me over here. First of all, it's, uh, exciting to be as part of this conversation. And frankly, you know, it feels like it's one of those days, 20 years ago when we were talking about how cloud was changing and what was happening on the cloud side of things.

It feels like the similar kind of transition, except instead of taking 20 years to get to where cloud got to, I think that entire 20 years has been compressed down to maybe two. And, uh, the pace at which that adoption is happening is also very interesting. I'll give you a quick anecdote.

I was talking to a CIO recently and, um, we were talking about exact same transition around, uh, cloud and how it, what it meant. And this said, you know, what, cloud was interesting because the benefits of that were more behind the scenes. It was more about the ROI that you could drive.

It was more about availability, it was more about resiliency, those kinds of things. And therefore, it took some time for people to really recognize that they could create these quote unquote multi-tenant applications and have it available everywhere, have economies of scale, be able to get better ROI over time. So it took some time to get there because AI was one of those things that it, it literally had a ui.

Like you could not miss it if you are talking to an agent. If you're talking to a chat GPT interface and you get all the answers that you were looking to get earlier, took you hours to do it, now you can get it in minutes. It's real, it's in front of you.

So it's not back office or it's not somewhere behind the scenes, it's right in front of you. So I think that inflection, uh, has become so much faster because of that reason. Uh, 'cause anybody can get it.

Anybody can understand that. And as a result of it, what we are seeing is that yes, there was a lot of experimentation that happened over the last two years or so. Uh, we're now seeing real tangible use cases.

There's still, I'd say, um, what you would consider as mundane use cases, but those mundane use cases over time, they all add up. So lemme give you an an example. It's a very interesting one.

I was talking to my team, which does simple things, product documentation, release notes, right? If you're in the software business, you do release notes. And when you do release notes, um, you typically go through a process where you have a writer who looks at the, uh, the, uh, to works with the product manager to try to see exactly what was developed, what was created, and you launch it, and then it goes through a vetting process.

It has the right tone, it looks at the different, uh, databases. All of that pulls that together. Takes, takes about five days or so roughly to write a good release note.

Okay? We have now, of course, introduced our own version of an agent, which is now able to take that entire five day process, squish it down to a day, one day. And, and now I think that one day is gonna go down to maybe an hour, uh, very soon.

And we do 200 release notes a quarter. So about 1,800 to a thousand release notes in year. And you multiply that time on what it's gonna save.

Small, trivial example, right? But it tells you the power of what this could be. Same thing with things like RFPs.

Our sales teams, every sales team anywhere they do RFPs, they respond to RFPs. If you're in the enterprise software business, you respond to RFPs. And, um, those RFPs typically take sometimes weeks, sometimes months to figure out how to, how to respond to that.

Because now you have an agent which can go look at a database. There's a rag pipeline built into the existing data stores for what we have. It has all the models built in, and it can go off, capture all the things that were answered for previous RFPs with a similar sounding set of questions, pull all of that.

And then now you suddenly have an RFP response, which sometimes took weeks now, literally in the matter of a couple of days. So it changes the game where you start looking at these small tangible, uh, use cases, uh, how you can create value for the, the work that happens on a daily basis. And I'm gonna keep going on, right?

But now it's, it's, it's, it's real, it's tangible. It's not anymore an experiment where it's about creating the best model that you can think of with the best accuracy rate, with the best latency. I think we've done that.

And now good enough is getting to good enough. And now it's about truly creating the ROI for these sets of use cases. So hopefully that sort of gives you a little bit of flavor of how this is moving and, and what we are seeing It really does.

And these are the types of conversations that are the most important. Right? I mean, I love technology.

com bust, uh, cloud, local, mobile, social, uh, pretty much, pretty much everything. And this is the, the biggest force multiplier that, that I've seen, uh, in, in productivity. I mean, the good news is we're debating, is it, is it a 10 x multiplier?

Is it a, is it a hundred x multiplier? Uh, and a lot of the research that that we're doing is, is, is around this, this concept around the digital knowledge worker where it's funny if you even talk to the h you know, uh, HCM vendors, right? They're talking about giving them an employee badge some sort, right?

But I'm curious, what does it mean, what does concept mean to you and, and OpenText and, and how does it changed the role of Ann the workplace? Yeah. I've heard discussions that were going from roles, uh, to workflows, right?

Right. So, interesting. Right?

I mean, let's talk, let's take a, take a step back for a second and talk first principles. So what is an agent? I think that question has been debated quite hot, uh, hot and heavy over the last, uh, 12, 24 months.

And if you go look at just the English dictionary of what the word agent means, it's something which has some agency. That's it. That's what it call boils down to.

It has agency to do something on its own. That's what a word agent means. So that agent can do a bunch of things which, uh, which it can do autonomously, or it can do things where you have rules, guidelines, and you've given them the authority to go do that.

But ultimately it's the agency that you have for them. And what is it? It's a piece of software.

And the piece of software in this case happens to be something that is taking data from a publicly available LLM. It's probably taking data, some from some private databases, which are there within the companies to make it a much more accurate representation of the answers. Um, and then there's some guardrails and rules, which are all written in a, you know, no code type of an interface, whether you're using auto gen, whether you're using link chain, whether you're using, uh, crew or other things, right?

It's all kind of built into that flow. That's the new AI stack that's emerging. So how do you then create it and then scale it one agent?

Fine, you can kind of figure out a way to, to force fit that within the workflow. But when you start talking about armies of agents where we believe that every single, uh, knowledge worker, you and I will have an army of agents working for us on a regular basis right? When we come to work.

So how do you create that, um, that synchronicity that orchestra, if you may, of these agents so they're not competing and they're also, uh, have access to the data that they should. So think about the world of IAM and roles and permissions that happened 20 years ago when, uh, identity and access management became a thing where now each of us had our own roles, permissions, access control, um, behind the scenes. Of course, it was an active directory in LDAP or something like that.

And then there were granular permissions about what data you have access to within the enterprise based on your role. That exact same thing is gonna happen with the agents too. So it's not too farfetched when you think about the fact that those agents will have their own employee id.

Um, I know it's, it's a, it's provocative statement, but it is somewhat true that they will have an identity which will have access to certain databases with certain profiles, with certain, uh, uh, uh, rules that they can go only, uh, access this data stores. And then over time those might change because they might retire because they're not doing as effective work. So they might have to get retired.

And then those, they're new agents that take, take their place and then so on and so forth. We call it the secure runtime for the digital knowledge worker. And we think that the secure runtime for this digital knowledge worker is probably going to be one of the hardest problems to solve for companies like us and others in our space to scale it so that it's available for mission critical work, right?

We are serving mission critical companies for mission critical use cases. And that's what you need to make it real. So, so, so Snet, just as a, a quick follow up because I, I want to pivot a little bit to risk in a moment, but I'm, if I'm hearing you right, it sounds like you see a strong augmentation with agents versus maybe a replacement strategy.

'cause there's a lot of debate on that. Yeah, it is. And um, I think, look, there's going to be a, um, a shift that happens to start for starting with augmentation.

That's the first thing. Um, you, you look, history's a good teacher. So if you go all the way back into the early 19 hundreds, the first time the, um, uh, the automobile kinda assembly line process started, and we went from being an agro economy to more an industrial economy, the exact same questions were being asked that, Hey, listen, I I'm working in the fields now.

It's all, all the jobs are going on the assembly line. Um, am I gonna lose my job? And, uh, what's gonna happen?

Well, they, some people did, of course, but then over time there was re-skilling that happened as well. So yes, there is going to be augmentation to begin with, but then over time I do see, uh, areas which are going to be more autonomous, and the people who are working in those areas, they will be re-skilling themselves to enable those autonomous workflows. So yeah, I, I do see, I do see a path, uh, on this, Dan, but, uh, right now I'd say the most immediate thing is augmentation followed by complete autonomous.

Yeah, I think that's a, a really nice way to describe it. You know, personally, I do believe there will be some real displacement. It's going to be upon us as humans to continue to invest in ourselves, to grow, to use the tools and become better.

And by the way, for everyone out there in the audience, you know, if you, if you listen to the recent, the one of our other day openers with, uh, Aaron Levy from Box, we, we talked a lot about this with Aaron as well. He sort of sees this too, about how economics grow, how productivity grows, how roles change. But it's a little bit of a scary time.

And speaking of a little bit of a, you know, it's a little scary just 'cause we don't know a lot. It's not scary 'cause it's bad. It's scary 'cause there's just so many unknowns.

And, and, you know, another unknown is really about risk. And, you know, the whole thing about trust, you talked about the, you know, you talked about that kind of operating system and what they get access to and what agents don't get access to. But with all this, the faster we move, the more risk we create the private data, finding its way to the wrong place is, is a big problem.

Sure. I'd love to get your take on this about how companies should think about being responsible with AI at scale. And maybe just any thoughts on how to build a framework for that and what you are learning in that, in that area?

Yeah, Great. Great question. Because, um, goes back to the notion of compliance and security.

Those things will not change at all. If anything, they'll become even more important as we go into this new world of, uh, both agents as well as digital knowledge workers and human knowledge workers together. I think it's gonna become even more critical.

And, uh, when the cloud, again, I'm gonna go back to that kind of metaphor. When the whole cloud, uh, change happened, it gave rise to this concept of a chief risk officer, a chief compliance officer, which became even more important because now your data was spread out in multiple places and you, you sometimes didn't have any visibility into it. So how do you keep track of that?

How do you make sure it's regulated? Um, I think you multiply that problem a hundred times, maybe a thousand times, because now it's not only about where the data is hosted, it's about how it's being used and who is using it. Humans Sure.

You know, they do. They, they make mistakes, they create issues, but you can track 'em. But once you have agents, then how do you do that?

Is there an audit trail? Is there something which has a timestamp for every single action that was done by the agent so you can track exactly what data they had access to, or what they didn't. Um, so I think there is a framework, there's some sort of a new concept of this sort of compliance and security for the agent workforce that will need to be created.

We are working on it internally ourselves as well. And we believe a standard, a more broader standard that can be applied across the industry is probably gonna be a good way to think about it. And, but I'm talking to three of our partners about this exact same situation as well.

'cause, uh, without that, the concept of trust won't exist with our customers. They might hear one thing from us, they might hear another thing from some of our other, uh, vendors, and you don't want that. Uh, trust and compliance is one of those things.

It needs to have a common language. So I do believe there's an opportunity to really create some sort of framework to enable this trust and security, not just for the world of cloud, but also for the world of ai. Yeah.

So Venet, uh, it seems like we can't go a single day without a new generative AI tool, uh, coming out. I can tell you it has kept, uh, kept my analysts very, very busy. Uh, it's good for analysts, by the way.

Sure, Of course. Uh, And, and we're using 'em on the consumer side. Mm-hmm.

I mean, and whether it's, whether it's search, whether it's, uh, deep research on, uh, on chat, GPT, uh, you know, I use summaries, uh, every, every single day. But doing it in a consumer fashion or a small business fashion is very, very different from, uh, a, a larger enterprise applying AI across, across business. Mm-hmm.

Data, um, a lot of theories on what the next frontier is. Uh, I've talked to a lot of different, uh, CIOs, uh, and groups and what they're doing. But what are you seeing as, as this next, uh, frontier?

Yeah. I think again, goes back to some of the use cases, right? Ultimately it boils down to, uh, where you see value.

I'll give you one example of the kind of things which none of us would've thought about, but it's, uh, it's a great example of like, duh, of course it makes sense. Testing, uh, develop developer testing. There are so many companies out there, whether they're tech companies or they are financial services companies, or retail or healthcare.

They all have it arms. They all have, uh, developers that are working to create software for themselves that they can use, right? Regardless of who you are.

When you do that, you te you test a piece of software. And when you are in a highly regulated space like fi, finance or healthcare, not only you need to test it, you need to keep an audit trail for all of those tests. Why?

Because when the auditor comes in, you wanna be able to show that this is exactly what happened. Right? Now, there are literally armies of people, manual people who are keeping track of all of those tests and putting up spreadsheets, which can be tracking when this test was done, when it passed, did it not pass, and where if it passed, then what situa which, uh, timestamp it passed, all that kind of stuff, right?

That's all happening. So people are investing in that. Now, here comes a very simple, elegant solution where, uh, where an agent can come in, literally take all of the data stores for where the tests are happening, uh, get some training based on that data, and then is able to authenticate exactly what tests are passing or failing based on prior databases, right?

So that changes, again, the game of months and perhaps weeks of time down to a couple of hours. That's mission critical. That's compliance work.

Otherwise they won't pass the audit. And there are some penalties associated with that. So I think this is a good example of where the same concept of the same model underneath, which was being used for a consumer use case can now then be used on the other side for a much more mission critical enterprise type of use case.

Yeah. It's actually a very novel, uh, a novel one. You know, you think you've heard all of these use cases and then you hear a new one Yeah.

And think, oh, I, I hadn't thought about that one. But that this makes, this makes perfect sense. There's a lot of data mm-hmm.

That, that goes into it. And these new technologies work better, uh, with the more data that they have. Very interesting.

Uh, I, I've seen you talk about this idea of, of do and done Yeah. Related, uh, to applying AI in the enterprise. Uh, can you talk about Sure.

What that looks like and, and, uh, how, how, how does it fit into the context of real business workflows? Yeah, sure thing. So, so what's interesting is that whenever you have something which is as nebulous as, uh, gen ai, when it started to begin with, and it became even more confusing when people started to use it for their own purpose.

And they said, how is it gonna be applied to this? And then they had to distinguish between what an agentic workflow is, which was the buzzword in the market to what an actual, like, uh, AI is. What's the difference?

So when you had all these buzzwords flying around, you sometimes have to kind of dumb it down and make it super simple to understand what is it that we are dealing with. So our method or our framework was fairly simple. It was search and summarize are things that you would do on a regular basis.

You mentioned you kind of do that to, right, where you look at data and you summarize some of the commentary. You probably consume a bunch of, uh, uh, papers from a lot of places. And you now summarize that together into this.

I call that search and summarize. That's a everyday kind of AI use case where you need it, you, you have it, and now the models have become good enough for you to get those summaries and you're not, uh, waste too much time sounds great, but then you get to the next step, uh, where the agents are doing the work for you. And then it's not just they're doing the work for you, it's just done.

So you are completely hands off. And at that point, you know, it's called unattended agents or it's called autonomous agents. You can put a lot of buzzwords on that.

But ultimately it's about just do and done. And you don't have to do anything at all. You, your hands off the wheel completely, kind of like the, you know, the, the autopilot thing.

So it's really a framework, uh, pat that was helpful for us to, it inspire teams, not just within our own company. You know, 24,000 people when you're trying to inspire them to get to a certain destination, you kind of have to dumb it down, make it easy, and then, but also do the same thing with our partners and our customers. And it started to resonate quite a bit.

So that's, that's what it is. It sits fairly simple. Yeah.

I said you're, you're making way too much sense here, seven, eight, don't, you know, we love, we love our acronyms and we need this to be confusing for, for, for everybody. This do and done thing is just way, way too straightforward, pat. That's a, that's a way That you can try to, you know, drive more work for your analysts, right?

By, yeah. There you go. There you go.

Some More, a more Acronym to understand that. There you go. But, Uh, but let's, let's bring this all home.

The real enterprise environment. There's a reason that, you know, I've seen stats as much as 99% of enterprise data has not touched AI yet. Now we can debate exactly how much it is, but we've seen consumer and kind of these frontline tools be deployed very quickly.

New LLMs, new scale, new rollouts, you know, billions if not trillions of dollars of infrastructure being deployed, but the enterprise has not moved as quickly. Yep. Um, it's not that it doesn't wanna try CEOs boards, they all want to get behind this thing, but it's taking time.

And a lot of the reason it takes so much time, snet, is because the real environments, the real application environments, the real infrastructure environments of these companies, the edge, the cloud, the on-premises, all these things going on is complicated. So talk a little bit, take us home by talking a little bit about kind of your thoughts around flexibility, openness, cloud, hybrid on-prem, all these things being addressed. How important is it for enterprises to, to understand and have the right kind of choice in how and where they run, deploy, and implement and build a sustainable AI strategy?

We can go days on that answer, to be honest with You. That's what I was gonna say. That's a big question.

Yeah. So it's a big question, but, uh, you know, I know in, in, in, in just a few minutes, you're gonna, you're gonna crush it with a great answer. So, And the bottom line is, look, we are smack in the middle of this, right?

Because, uh, if we have 120,000 customers, most of the Fortune 500 of our customers, so we talk to them, I talk to them on a kind of regular basis to understand how they're thinking about the data. Um, what I do know for a fact is we have data sets that we have from our customer bases across all of our product lines that I know if we were to, um, extract it and do the right kind of ETL on it to, uh, simplify it, to make it secure and then be able to use it in a way that we are training and, uh, implementing some of the more rag pipelines on existing LLMs, then a hybrid model, which takes the publicly available data with the data that we have inside the firewalls, behind the firewalls, will create a much more accurate representation of the answers by the agents and make it a lot more productive for our customers. But, but having said all that, it's a really difficult problem to solve at scale.

Why? Because you cannot make a mistake on the privacy and the security of any of those data sets. It's not a technical problem anymore.

Yeah, you can take large data sets, you can do ETL on it. You can create the right data pipelines, you can create the right data, uh, uh, preservation, the retention models, all that stuff. It's been solved.

We know how to do that. But it's about then how do you make sure that none of that leaks over so that you're trying to create the model for a one customer and kind of leaks over to the other customer either. So it's, it's a, it's a, I think it's a foundation of what needs to happen over time, and we are working on it.

Uh, I know others are working on it as well, and it's going to create the right accuracy and confidence and trust from the customers as well. But we are just scratching the surface on making it scalable, uh, and doing it in a secure and trusted way. Yeah, I think, I think, uh, first of all, I want to just thank you so much, SNET, because you brought a lot of pragmatism to this conversation.

You know, Pat's a, Pat's a former product guy that used to have a real job. I'm just starting to say the things he used to say. Uh, and now, now he's an analyst and he likes to say that.

But there is so much of this kind of, we talk in these grandiose kind of high ways. Um, you had me all pumped up about those agents that are going to just work 24 7. I, I actually, I really love it, but there's a lot of work for us enterprises that have important proprietary customer data to get from where we are to where this can work for us every day.

This was an absolutely wonderful way to kick off day four here at our six five summit. I wanna thank you so much and hope you'll be back with us, uh, for some more conversations. Savix, I wanna track the journey.

I'm sure there's gonna be a lot of good stuff happening. Let's do it again soon. Looking forward to it.

Thanks for the conversation. Thank You to everybody out there for joining us for this day four. That was a great opener.

Great way to start. Talk about unleashing ai very practical, real stories from both their customer Zero, what they're doing inside OpenText to how they're helping and partnering with their customers. com slash summit.

More compelling content ahead. Stick with us. Hi everyone, and welcome to the six five Summit AI unleashed for this enterprise app Spotlight.

I'm joined by Chris Leon, executive vice president for Oracle's application development on how AI agents are shaping the future of organizations. Thanks for joining us, Chris. Pleasure to be here.

Mel. Good to see you. Good to see you.

So Oracle has introduced a range of AI agents across its applications. Can you share a specific example of how Oracle AI agents have improved a key process for employees or managers, perhaps in HR or supply chain, and what measurable outcomes have resulted? Sure.

Um, you know, good, good, good news is we've kind of started with a whatever call, crawl, walk, run kind of approach. Um, we've been delivering generative AI services for the last, you know, 16, 15, 16 months. Um, where we started, what, what I would call kind of the, the crawl phase where we were helping, um, just kind of automate basic processes for, for some of our customers.

And I'll give you a few examples of those, um, uh, processes like, um, pre-building a job requisition based on, um, the user's role and a little bit of background on what assignment they're trying to fill, what role they're trying to fill, and being able to take maybe, um, a job requisition and created in a matter of minutes versus maybe a half an hour to 40 minutes for, for a, uh, manager to create a really compelling job requisition. So if you think about, um, just that example, if you know you're doing 5, 6, 7, 10,000 job requisitions a year and you're saving 30 minutes per job requisition, um, that starts to add up to, to real dollars. And, and I'll give you another one.

Um, as we moved kind of beyond the, the, what I would call, you know, generative AI services, um, work that we've been doing into kind of more the real AI agents and, and as, as we define AI agents, it's, it's really kind of using the large language model as the brain, the decision maker as to which tool they need to call in order for them to make the right decision. We started building a lot of AI agents across sup, supply chain management, hr, cx, ERP. Um, so I'll give you a a great one.

Um, we, we built a, um, maintenance repair advisor agent. And so what this agent is able to do, and, and, and how it helps is, um, a maintenance work order, um, may get generated from a connected equipment. A machine goes down and it generates a maintenance work order.

And what usually takes a maintenance technician time, um, to evaluate, understand what, what the issue is, maybe look up an error code, maybe look up the service history of this particular machine, and then identify, um, who is the right service technician to be able to, to fix that particular, um, issue. Instead of doing that in a kind of sequential step-by-step manual process, we created a repair advisor, a technician repair advisor that can do all of that virtually. It has access to, um, the work order, the maintenance work order, it comes in, it can look at the machine, it can look at the service history, it can understand what the error cones mean from the manual that we've stuck into our vector store.

Um, it can identify potential fixes, um, and then it can dispatch a service technician to go fix that particular, uh, machine. And so what this can do is lead to better, more uptime, um, for these particular machines. And, and as you know, the more uptime you have, the, the, the, the less costs you have because your machines are running longer and, and, um, you're able to, to save the number of steps.

You don't have to have a person engaged every step of the way. So you can save dollars, you can keep the machines up longer, and it can lead to real, real dollar savings. I appreciate you defining what AI agent means to you, because I think throughout the industry that sort of varies and so that it's really helpful to kind of say like, well, this for us, at least for now, right?

Yeah. Because I think this could possibly change. That's what that means.

So you also, Oracle also allows, in AI Agent Studio, you allow for your customers to tailor these AI agents for their unique needs. So how do you see customers using this capability and how do you see that shaping the next wave of enterprise enter automation? Yeah, no, Really, really good question.

So, um, first as I said, the kind of the crawl, walk, run, um, scenario, you know, we started building generative services, AI agents, rag based AI agents. Were, were using our document store in, in our Vector database, um, for the last 16 months, as I said. Um, and then, um, we, we came to the conclusion that we were using, um, an internal set of tools that my development organization, the entire Fusion application development organization was using, um, to build these agents and associate them to business objects and APIs in, in our, um, fusion application suite.

And so we decided to turn that over to our customers and partners in the form of what I'll call an ai, um, IDE or a development environment, and that's called AI Studio. So AI Studio is really that ability for not only us to build brand new AI agents, seed them and let our customers configure them and use them, but really allows our partners and our customers to extend what we have or to build agents from scratch. And so, so to answer your question specifically is, um, I think our customers will start with, hey, taking that maintenance tech advisor agent loading up their, their documentation, um, and maybe at the end sending, you know, adding a step to send a summary at the end of the day, um, to their manager about all, all the, you know, maintenance work orders that they may have fixed.

So they might take what we have 'cause they're just kind of learning and extend it, but partners will be able to create, um, very, very verticalized industry specific agents because they have a lot of domain knowledge, they have a lot of understanding of what our customers are looking to do, and they can just use their imagination and, and the power of what, you know, agents can really do for, for business. They can really automate, um, these processes in ways that we weren't able to do before because they can think on their own, right, they can be more, um, probabilistic versus just kind of the way we've always coded everything, which is step by step by step. And we know how it's always gonna end.

Now we can let the agent make decisions about what is the right tool for it to use to solve a particular problem or, or a particular use case. And so I think our partners are really gonna be able to really just imagine new types of solutions and really fill what I would call some cracks in, in our processes, maybe even to extend to systems outside of our application and bringing external data sources in order to make the solution that they develop even more robust. And as agents sort of start to take on some of this more routine work, what are you, how are you seeing kind of the, the more urgent skills gaps that are emerging for enterprise teams?

And and how do you think that organizations should start preparing their people for that today? Another, yeah, it's a really good question. You know, I think, and what, what I've seen, um, Mike, and we just came off of a very, very large internal hackathon where we had, you know, um, large part of our development organization really experimenting and building some pretty cool, um, agents.

Um, some super unique I would never thought of of those. And, um, but, um, the skill that I think they all, um, first had to understand is what is the unique power of building an AI agent? What can they bring that we couldn't do with some of the technology of the, of the past?

What new additional sources of information, new ways to automate new decisioning that these agents problems that they can solve that we couldn't solve in the past? And so that was the first kind of hurdle they had to get, get by. And once you start to understand, look, I can bring in best practices information, compliance information, marry that with data in our system, I can marry that with internet specific data.

I can look at the weather, I can, you know, I can look at, you know, uh, internet searches and bring back different types of information and start to solve problems and not just recommend, but actually take action in our system. So I think the first problem that people need to understand is what is the types of solutions that you can now solve that we couldn't in the past? And then I think the second thing is how to really get your mind around developing these system prompts in order to, to, to really, um, ask, you know, or tell the agent what it needs to do.

So it's really common language, but there's ways to write these prompts that, um, are, um, very specific to what you want it to do. And there's different ways to go about it. You can start with a very, very complex prompt.

They don't recommend that, or you can start with kind of trial and error. Start with a basic prompt, see if it does, you know, solves this solution that you're asking it to, to do, call the right tools if it doesn't add another, you know, step in that prompt. So I think they really need to get their mind around how this common language, English, or whatever language you speak, can drive the behavior of, of these agents.

And I think those, those two skills are, are things that people need to develop. Yeah, it's interesting, this like digital workforce, people are gonna have to start to learn to work with kind of like a new teammate as Oracle is enabling more autonomous AI driven workflows, also having to deal with sort of building this trust between people and AI agents. And that could potentially prove harder to, than deploying the technology itself.

So where are you seeing kind of that start stop mistakes in organizations, how they can avoid that? What advice do you have on that? Yeah, you know, I, so what I tell my team is, you know, people are worried of, you know, are agents gonna replace people?

How is that gonna work? You know, what I, what I tell my team is we need to take advantage of this technology. Everybody should be at least a five x employee, maybe even a 10 x employee, leveraging the partnership that we can have with, with ai, whether it's agents or, or just generative AI services.

We all need to be more efficient and more productive in our day-to-day work. So, so that, so that, that, that has to be kind of, kind of first and foremost. Um, and then what I would, I would tell them as far as, um, where they can, you know, where, where, where you know, kind of agents will, will fit in and how you can make sure that you're getting the right answer.

And the accuracy rate is high enough, is first define what that needs, what that needs to be. Um, is the, um, error rate 95%, is that gonna be okay for this particular problem? And if it's not, can I stick a human in the loop?

And, and we have those capabilities where, hey, look, before I, you know, charge your credit card before I place this order, before I send out this service technician, do I send it to the supervisor and have him say, yes, this is the right thing to do. So I start to automate these processes kind of one step at a time until I get more comfortable. And then maybe I can take the human outta the loop and that step, take the human outta the loop and that step and then becomes a fully autonomous process.

But the idea and what people have to think about is, this is not about job elimination, this is about productivity. Um, the ability to increase productivity for organizations. So you will start to see small organizations, medium sized organizations, being able to be as productive as very, very large organizations because they're gonna be able to take advantage of these autonomous processes.

And you can do it step by step, as I said, pulling a human out of the loop to make sure that you're more, more comfortable. I'll give you one more example, it's kind of short one. Um, we have, we rolled out a benefits, uh, advisor solution to, you know, 60, 70,000 people in across our Europe, um, organization.

And it answered questions on benefits or different benefit plans, compensation plans. And, um, the, the SVP of HR said, you know, his comment back was, the regulations in Europe for, for, for some of these benefit plans is very, very complex. And even their best benefits advisor has a 85 to 90% accuracy rate.

So if we're at 90, 95% accuracy rate with an eng with an agent, that's better than where we would be without an agent. You know, so, so you have to take into account what is the accuracy rate and where you want to stick human in the loop in order for you to feel confident in, in the solutions and the answers that you're getting. Yeah.

So you mentioned like people should be five x 10 x, but then there's, you can take it kind of a step further with like, the value of integrating HR supply chain data with these AI agents. So how does Oracle's approach to that cross-functional integration help organizations become even more agile and resilient? Yeah, you know, I think, I think what it, it, it kind of breaks down the, the boundaries.

Uh, and not that we couldn't have done this before, it just makes it easier to grab different pieces of information to help drive decisioning in one part of, of the system. Um, and, and I'll give you, uh, I'll give you the, the skills example in, um, in the technician repair example. So skills are something we traditionally have in hr.

It's a very, very well organized HR process, not something we have traditionally done in our maintenance application, right? It was very rudimentary. We had a few skills.

Now I can easily grab that talent information, that skill information and go look at what this technician has and be more accurate, oh, they've worked on this type of C and C machine and they've done it for this many years, and they can be come to the top of the barrel. Versus in the past it was CNC expert and you just stuck them in and maybe they didn't know this particular machine, or they didn't know how to fix the particular problem. So now we can solve these kind of broader problems because we've kind of made it easier to open up the information that we can grab from different parts of our system and even external parts of our system.

So we can go outside, like I said, into the internet, into services outside, and bring in external information to help make better decisions inside our, our organization. Yeah. So in a way that's not just, you know, the sort of easing the fears of AI taking your job, it's actually AI is helping to help people like advance in their careers.

It's putting them in the right place at the right time and kind of figuring out where people belong. I think that's, that's actually where, yeah, I, I think, I think, you know, where, what it's gonna come down to in the next year, six months, nine months, is organizations are gonna have this kind of productivity dial that they're able to turn up and, and it's gonna give them more advantages to say, Hey, look, you know what? I can deploy.

Maybe, maybe I don't need to deploy as many people in this one area. I can turn on a few more agents over here and take some of the people and have them work on this, you know, more cognitive oriented problem that I really need a different type of leader to, to be able to solve. And so they'll be able to, you know, it's kind of elasticity in our cloud.

They can turn up elasticity to productivity over here and, you know, move resources over here and give a lot more flexibility to the problems that they can solve, allowing them to solve more problems and not, you know, always having to be the largest organization that can compete in, you know, different industries. So if you had to make one bold prediction about how AI will reshape enterprise organizations over the next, say five years, what would it be and what's the most important action that you think leaders should take now? To, to be honest, I, I think in the next five weeks, maybe five months, you know, so it's really hard to think five years out.

'cause it really is, things are changing. Yeah. Things are, things are truly changing that fast.

You know what, what I tell my leaders, and I, I literally just sent out a note to all of them. Um, they need to look at a few of their key deliverables that they have to their customer, internal customer that they support, and they need to automate those. Um, a couple of those by the end of this fiscal year or this, the end of this calendar year.

And I said, you need to start thinking about, hey, look, these are deliverables that we have, these are problems that AI can solve for us in a much more automated way. Let's go automate those and get to 80, 90% automation and then go to the next step. I'll give you the one example I gave to my strategy team that helps, um, enable our field.

I said, there's different ways to learn now, it's not a PowerPoint presentation and a webcast that you go have. Um, these, these, um, AI can create podcasts and they can create these dynamic podcasts where two people are talking and interacting just like we're interacting from a document or from a PowerPoint. And it's different ways to learn.

So we, we have to be smarter about how we can deploy these new technologies so people can learn easier. This is how kids are learning in college, right? They go to these websites, they take notes, and they get these flashcards and they're still learning the content, but we've made it much more efficient for them to do it.

And that's how we need to think about different parts of our organization that we can make more productive. So I've challenged each part of my organization to say, Hey, look, you need to think of processes that can be more automated and take advantage of the pattern recognition and, and what these large language models do really well, and let's automate those processes. That's awesome.

I love that. I tell my kids all the time, use, uh, use all the AI tools that you can because that will give you the superpowers that you need when you join the workforce. Absolutely.

Well, Chris, thank you so much for joining us and for everybody who has tuned in, thank you for joining us for this enterprise app spotlight at the six five Summit. Stay connected with us on social and explore more conversations at six five Media slash Summit. On behalf of six five Media, thanks again.

Hackers Plant Raspberry Pie in Bank Network. Ed brings AI observability and Acura acquire scale computing. Cisco donates their agency and sneer wants faster smarter AI psycho uses AI for DevSecOps and Trumps new tariffs, disrupt industries, and raise tariffs.

All this and more, maybe some snam all on the tech field day rundown. Welcome to the Tech Field Day rundown, where we, each time we meet, we run down the IT news of the week with a variable degree of snarkiness. I'm your host Alistair Cook, and I would like to welcome you to Root Beer Float Day.

Uh, I'm also welcoming for the first time to the tech field day rundown, my friend Mr. Jeffrey Powers. How are you doing today, Jeff?

Is is rip your float day in New Zealand or here in the United States? Well, I think it's there in the United States along with national wiggle Your toes day and wig your toes is important. If you're flying from here in New Zealand up to the United States, you gotta keep the blood flow going.

Absolutely. Blood flow's very important. com for enterprise build needs.

It is also National Social Engineering Day, or as I like to call it, national Kevin Mitnick Day, the master of Social Engineering Love Kevin Mitnick. We start with, uh, an interesting story. We're a, uh, cyber robbery group known as UNC 2 8 9 1, that's a catchy name.

He used a raspberry pie device to an infiltrated bank's ATM network. The device was installed in an unmanned ATM location, and the hackers initially used a cellular connection to initiate their attack. And it seems that their plan was to gain control of the ATM network and extract cash.

Hmm. Modern Day Bank. Rob is a sophisticated, don't need guns or masks, do they?

Jeff? No, they do not need guns or masks. You know, it's really funny because every time that I go to a pump, I go to any type of a TMI insert my anything that I have to insert my card or tap my card and run my hand across it and see if that thing just pulls off.

Uh, if it doesn't have a protection sticker on it, I do not pump the gas. Same thing with ATMs and even cashless registers that are in the store, because you, you might think that they're safe from skimmers, but you have all these report reports that, that scammers are actually getting inside the store, maybe through a disgruntled employee that, you know, getting a couple dollars on the side turns the cheek, whatever. So, and, and I always have to ask people, when was the last time you used a McDonald's kiosk and have you ever thought of checking those reader card readers before you even started, started the card?

Now this news story is the greatest reminder that even in enterprise it, especially if you're in the financial enterprise, IT area risks come in, uh, from physical vandalism and physical hacking. It's not just enough to secure the software. You need a full stack physical security, access control and monitoring.

All invol involved in this and the threat threat vector, threat vector we have here wasn't a bug, but it's a box the size of a wallet for IT. Pros, this is the kind of breach that slips past traditional defenses. And it's time to actually start thinking about planning different types of hardware and software layers to make sure that if a device is planted, your credit card or other personal information is safe.

And to top it off with a good, good security system that can identify maybe, maybe through AI and, and that being physically altered ha that event happening so they can, police can track 'em down a sensor inside the machine that could identify weight or difference, uh, card slide slides in a little bit differently, all of a sudden could create an instant popup on those screens saying, Hey, this kiosk has been compromised. And finally, a system that identifies possible risks and ways to record every bit of information. So when a kiosk is compromised, law enforcement have enough data to find the person that did it.

And that can all happen at the server level. And I think that that's where we need to see, uh, you know, new stories like this and how we have to counter them. Riverbed has launched new AI powered tools to improve network monitoring, including the Intelligent Network, observability Essentials Bundle, and the XX 90 appliances.

Uh, these updates help businesses manage growing data and the cyber threats using AI to spot and fix issues faster. Key features include tools for visual visualizing, network activity, AI driven alerts, and better support for cloud and on-prem setups. The new XX 90 hardware boosts speed and storage for analyzing network traffic.

Riverbed has also introduced a flexible subscription plan and says its observability sales grew 92% in early 2025 Alister. Do you think that's true? 92%?

It's a huge amount of growth, but it depends where you started from. If your observability practice was small and just beginning, your growth can be huge. We know that over time it takes a while to move those needles, but I think it, it's interesting to see Riverbed adding more and more observability, uh, seeing what's happening in real time and getting that feedback loop back to application developers and operations teams is vital.

And we've always seen, uh, Riverbed as, uh, about application delivery management, well observability as an extension of that to seeing far deeper into the applications themselves. So this brings together a couple of things. Uh, I really like that there's a plugin for Grafana, which is a, an enabler for creating custom dashboards and, uh, getting open source, uh, tools for visibility in your environment.

That's, it's really nice to see that they're not locking you into using just their own tooling. Uh, having an AI tool that looks at what's going on. It'd be interesting to see how that AI tool plays out, how national AI it is and whether it is more of a exception handling and identification, which would be more of the predictive AI that we've seen historically in management tools or whether there's something more in the, the generative ai.

And I really would need to dig into a little deeper into that. Hopefully we'll see Riverbed at an upcoming Tech Field Day event, showing us some of the depths of this, and particularly showing us some of these amazing, uh, high performance devices with sustained packet capture of over 50 gigabits per second being dumped out to, up to two petabytes of storage in these XX 90, uh, appliances. So there's some really interesting capabilities in these.

Aira has acquired scale computing and is retaining the scale computing name for the combined entity. Uh, aura's Secure Edge networking joins scale computing's virtualization technology, and they plan to be a major disruptor in the edge computing, networking, helping businesses run AI and other applications across many locations. This move aims to simplify operations, improving performance, and uh, serve industries like retail, manufacturing and hospitality.

The former, uh, CEO of scale computing Jeff Reddy will serve as the president and chief marketing officer and the company explain plans to expand its global reach with smarter and more efficient edge solutions. Uh, this pairing together and staying under the branding of scale computing looks like a really cool move, doesn't it, Jeff? Uh, well, we'll see what happens on that.

Uh, you know, I I think they're just kinda saying, Hey, uh, scale computing's in a move where we're bas they're basically shouting. We swear Edge, edge isn't buzzwords in boxes anymore. And sure if you, if you're an IT manager that has like a hundred thousand fast food locations, this type of marriage might mean smoother ORCA or orchestration, uh, real time AI at the edge and one less reason to call your SD WAN vendor.

But, uh, just kind of tell 'em that, you know, the shake machine's still down getting more real. Let's kind of pretend that this is all plug and play magic, which it might, it's still edge. That means, uh, remote boots means compliance.

We've got troubleshooting boxes. Uh, you'll have to troubleshoot in Omaha, Seattle, and Miami from your bed at 2:00 AM in the morning with the only screen on being your laptop screen. I'm bright, but you know, hey, now comes with an all-in-one interface with prettier dashboards and a bigger sales team.

Uh, on the security front, it might solidify them as being the absolute best in cyber security for edge computing in a distributed enterprise, but only time will tell if they can, uh, take that belt home with them. Agency. A project that helps AI agents work together across different systems is now part of the Linux Foundation, backed by companies like Cisco, Dell, Google Cloud, Oracle, and Red Hat.

As more platforms create their own tools, agency agency aims to unify them with open standards for discovering, identifying messaging and monitoring agents so they can easily collaborate like humans do. By donating the full framework to the Linux Foundation Agency ensures that the project stays open, it stays neutral and community driven, helping building trust foundation for agent, uh, based AI systems across the industries. So Alistair, is this type of system better as an open source cross platform type thing, or did they make a mistake?

Well, I think it's definitely one of those situations where we're seeing a plethora of standards turning up in Ag Agent ai. So AG Agent AI is where some AI system is not just doing an analysis, but it's completing a task. And normally it's a small task and, and you build a larger system out of a series of small agents doing very specialized tasks.

The agency framework aims to make it easier to use agents that aren't just from a single vendor. So you don't have to commit yourself entirely to just one company's, uh, agent mechanisms. You can combine agents from different vendors.

Interesting that this came out of Cisco, uh, uh, was actually from out shift a, uh, a component of, uh, Cisco and, uh, supported by Dell and Google Cloud and Oracle and Red Hat. And so there's, there's a bunch of people behind agency. It's a related standard to the a to a standard that we've previously seen.

So a to a communication between agents, uh, and is sort of sits in parallel to the model context protocol MCP standard that we've seen as well. I think it's a sign that agents, uh, and agent AI are starting to mature when we get beyond, uh, a vendor who says, you have to come to us and get everything because we are the only ones with a vision. And we get to the more mature position of we'll have agents that are, are working on our technology, and they'll interact with agents that are working on other vendors technology.

I think that element of being able to build a multi-agent software lifecycle across different vendors is really important. And the only way that happens is through standards that aren't entirely owned by a single industry organization. So putting this framework out into the, uh, Linux Foundation, absolutely, it's a, a good move.

Time will tell which of the standards we're actually gonna use over time. We've seen this previously that there were, were a whole lot of different ways of doing orchestration of containers and eventually Kubernetes won out. We will see the same thing here eventually, the collection of standards or the single standard that unifies will emerge as as a popular acclaim sear, the storage networking Industry Association believes that CPUs could be a bottleneck between storage and GPUs and AI systems.

The storage AI initiative places the storage on the same ultraverse network as the GPUs, and it appears that the optimal infrastructure design for AI infrastructure is still very much a moving target. Jeff, where should I attach my storage in my AI infrastructure? Well, first of all, gazo tight.

Um, and that's a, that's a good question. Where should we, 'cause it looks like SNE has just kind of hit the panic button right there. Uh, hopefully it's gonna bring together industry leaders rethink how we feed that AI beast.

Uh, enterprises are always basically burning through power. They're burning through dollars, and they're burning, burning through IOPS because their infrastructure just wasn't built for that AI pipeline. So hopefully this project will look at aiming to create vendor neutral standards for AI optimized storage through high throughput, through low latency, through tiered data flow.

Um, for IT leaders, it's a future proofing move. If you're not storing data anymore, you're feeding intelligence engines that needs a, basically needs a whole new playbook, and that's where they need to start writing here. Um, I, I consider it to, like back in the day when we had to compress files to save disc space and transfer rates, but even using, I've even been using AI to knock down some of my storage needs, uh, running programs that find duplicate files on systems and kind of mixing 'em, finding out which one's the better one, uh, uh, finding systems that actually can be reduced.

Uh, so there is more storage, but then that means there's more storage to be had and more files to be had, and then we run into the same problem and we run into the same hardware limitations from there. So we will see what happens on this whole thing if it actually does help us in the future. Psycho has added an AI tool to a security pro platform that helps teams figure out which software issues are the most dangerous so they can fix them first obvious.

It also offers other AI to tools that suggest fixes and track code changes. These tools give better context and reduce false alarms as AI creates more coding mistakes and hackers to use AI to find them. Having smarter security tools is becoming more important with new rules and sharpened responsibility between developers and security teams.

Keeping software safe is now a bigger focus. Do you agree, Alistair? Well, I think there's an arms race going on between bad actors and, uh, software developers or the infrastructure and DevSecOps teams, and that's not news, it's just a change in the tooling that's being used.

So site code here is doing some analysis of changes that are being made in source code repositories and identifying things like vulnerabilities in those source code repositories. And more than just identifying that vulnerability exists, but looking at the criticality of that vulnerability, whether there are other mitigations for the base level of vulnerability built into the code, or particularly the infrastructure around it. Uh, the psycho tools are aiming to make sure that as developers are fixing vulnerabilities, they're fixing the ones that are actually a risk rather than just hitting everything that's got a CVE number assigned to it.

I think that's, that's a really important thing because focusing on where the risk lies is vital to making sure that your safest from compromise. Uh, this application security posture management platform that is from Psycho is definitely gonna be helpful, particularly if you're working with open source kind of projects as well. But anywhere that there might be a supply chain attack, and we've seen supply chain attacks on people's source code repositories.

We saw, uh, we covered in, in the rundown, I think it was last week, around a compromise that was in a plugin for, uh, visual Studio code that was allowing an attack through that Visual Studio Code plugin. These kinds of elements of, uh, code coming in from external sources needing to be analyzed and needing to be looked at the changes. This is exactly where site code is, uh, attacking with these AI agents, and we should see this being really helpful.

Um, future and group surveys found that these kinds of dev SecOps automation tools and cybersecurity automate orchestration are really important things for our customers, and that they really are looking for ways to help reduce the risks around the high speed and open development of applications that they have. It's time for us to take a little bit of a closer look, and it's a topic that's getting, uh, a lot of mainstream news coverage and is definitely impacting some of the organizations we care about. And it's President Trump's new tariffs on dozens of companies that are shaking global markets and hurting major US companies like Apple, Amazon, and car makers.

What's going on is that they're increasing the cost of materials and products that are coming in, and the uncertainties are leading to fairly significant delays in some of those things being shipped. 1 billion in tariffs just this quarter, and Ford estimates $2 billion for the entire year. That's money that's being paid by American consumers into the government's tax funding.

Uh, the tariffs do aim to fix the trade imbalance and the tariffs are being levied against, uh, cut foreign companies that sell more products to the United States than they buy from the states. Uh, and the objective is that this makes it more competitive for US-based companies to build things. And this takes a while.

Uh, there's our problem. There's a, a big lag time between having this tariffs come in and the jobs and the productivity on shore that the tariffs are supposed to encourage. Uh, right now job growth is slowing because we have that latency between, we have that delay of higher cost for products, but we don't yet have the ramped up production domestically, and it's likely to be some time for that.

Uh, experts warn that the economy could suffer as it slows down and more costs are passed on to consumers. And of course, AI rears its head that some people believe that this, uh, AI tools are gonna be used to eliminate the people cost and lead to, uh, reduction in, in employment. Now, historically, we've seen this panic that every new technology is going to take away jobs and yet don't have fewer jobs working in it.

We have more jobs caring and feeding those machines over time. So I think it's interesting to see that there's a, a large list of companies that are, uh, having tariffs levied against them. Here.

I I live in New Zealand, we have a 15% tariff as being levied against New Zealand because we send you more wine and, uh, cheese and, uh, dairy products or other, other meat products then we get from the United States. Uh, the president does say that tariffs are vital and that, uh, there's no chance for us survival or success without these, these tariffs. But, uh, I think he's a little uncomfortable with the fact that there is a lag between putting the tariffs in and, uh, and the actual upsurge of US manufacturing and employment.

And so this is why we saw a bit of a knee jerk reaction to the news that, uh, the rate of increase of, uh, employment in the United States has gone down. That's below the, uh, the average of 130,000 jobs a month while 73,000 jobs added in in July. And that wasn't welcome news.

Jeff, are you seeing confidence here? I'm not in the United States. I only see this secondhand.

Do, are you seeing confidence and, um, comfort in, in the United States as a result of these tariffs? Uh, before I answer that, I'm gonna have to, uh, invoke my 30% tariff that just happened. So, uh, once I get paid for that, then I can reply.

Okay, now we're good. Um, so I, this is, this is just, this is amazing how one person can think that doing something like this is going to help overall when it has been proven time and time again that it does not work. Uh, that going back to that broken wheel and trying to make a new version of that broken wheel is going to do it.

And that's, that's what really frustrates me because, uh, we're in the IT area. We see from everything from consumer products all the way up to enterprise products that things that are just, you know, we can't get to anymore and we have to have those global tech supply chains and enterprise it definitely feeling that ripple effect through the whole thing. Uh, I personally had some work put on hold because they don't know what it's basically gonna mean for technology and tariffs.

That means I'm not working, they're not working and nobody's winning off of this situation. So with companies like Apple expecting to pay that $1 billion in the quarter, it, the cost increases are just accounting lines. They're pressure points, not just account, uh, accounting lines.

They're pressure points across procurement, across infrastructure, rollouts, and end user pricing. And that means potential delays, as you said, in refresh cycles, uh, higher costs, everything. Uh, the next iPhone is gonna be what, 20 to 30% more.

And when Apple tried to move to a different country to relieve themselves on different taxes, we got dinged for that as well. And building, let's, let's talk for a second about building infrastructure in the United States to alleviate this whole thing. 'cause you just can't pop up a, you know, an Amazon store in the middle of, you know, Utah or something like that and expect production to happen with within under 12 months.

Because building a plant like that is more, more than just breaking ground and building the plant itself permits environmental surveys, uh, government getting in the way to try and figure out where their cut's gonna come in. And more than that, that's gonna take more than 12 months alone. So if we were to build a plant, 'cause we tried this, Foxconn came, was gonna come to Wisconsin and do all this, and they went through so much red tape.

They, they first started in Ohio, they went through red tape, they moved to Wisconsin, and then we ended up not getting Foxconn at all in the United States. So the, it, it just doesn't make sense. I have, it's set up for failure and that's what really frustrates me.

Adding in things like AI stressing infrastructure. That's the only option is to pay. And that means sending staff and sending these bumps to the client.

And that's what's really, I can see it. I don't understand why other people cannot, but I dec digress. I I think, you know, the, the, uh, the plan is, is a long-term plan.

It involves pain at the moment, and I think that pain is, is gonna take some time to pass, hopefully some more of the trade deals, uh, negotiations go through. And, uh, we see some reductions in these tariffs over time. But, uh, that's really comes down to the strength of those negotiations, the ability of those negotiating change.

Speaking of the real world, we have some things coming up. Uh, Jeff, you're gonna be at Cleveland for an event, uh, in the next couple of weeks? That is correct.

I am going, it's gonna be my first time at Share Cleveland. This is everything that deals with mainframe. IBM just came out with a new AI mainframe, uh, a couple months ago and we're gonna probably see a lot about that.

We're gonna be, uh, that's going to be of course a hotbed topic is bringing AI into mainframe. Uh, we did a podcast on this a couple weeks ago, uh, talking about share on, uh, tech Field Day. And, uh, we all thought, uh, thought the exact same thing that AI is going to be where it is, IBM's new servers are gonna be where it is and the new direction, uh, that, uh, mainframe is gonna go to keep people going, keep people secure.

And 'cause you know, we have the issues in New Jersey where getting a flight is just as painful as getting a tooth pulled, uh, seven times over. And so how is the new direction going to affect finance, going to affect flight, gonna affect trends, gonna affect buses and all that other stuff? Uh, so it, like I said, it's gonna be my first time, so I'm not exactly sure what I'm gonna expect from there, but we've got a full schedule of, uh, different clients that we're meeting and talking to and it'll be interesting to see where their direction is headed.

And then the next event for us after that is AI infrastructure Field Day. I'll be hitting back to Silicon Valley with my panel of delegates and what we're hearing from, uh, Broadcom and Hammer Space. Martis and Satir, uh, um, potentially a, a bunch of other interesting sponsors.

Also gonna talk about how you build an infrastructure to run all of that ai. Following on from that, later in September, uh, Tom Hollingsworth will be back talking all things security again in Silicon Valley. He'll have Nile and Square X and one password as presenting companies for that event.

And you can check out all of our Tech Field day events on the tech field day com website. Thanks for watching the Tech Field Day rundown with all of the news that's ready to be run. You can catch new episodes every Wednesday as a YouTube video or on your favorite podcast application.

The rundown is streamed on text, on TV as well. And you can often catch us at other Textron and RUM group events and programs. We'll be back next Wednesday to talk about all the IT news of the week.

Uh, and until then for myself and for Jeffrey Powers and all of us here at Tech Field Day is wishing you and yours a great day.