Techstrong TV October 6, 2025

Is AI in danger of becoming the biggest paper tiger there ever was? I don't think so. You're watching Text Drunk Gang.

Hi everyone, it's Alan Shimel for Textron Gang. Thanks for joining us. Um, you know, as you may or may not know, we, we usually record the Textron Gang one day before it plays.

So we try to be as current as possible, but no one wants to come in and record on Sundays. So usually Monday shows are recorded on Friday, and since it's, it's Friday, I've gotta take a moment to bask in the glory, right? There's nothing like sending the Red Sox home empty handed.

I, I gotta hand it to this kid. Cam Sch Schlater, who's a, a Boston kid, a Massachusetts kid, by the way. But man, how sweet it was.

Mike, I, I know you weren't drowning in your beer. No, it was sad. But, you know, my oldest son is a diehard Red Sox fan, so I think it's probably gonna be 72 hours before I hear from him As, as it should be.

Anyway, a number about baseball, though, we've got a lot to talk about today. Actually, before I do, we should give a shout out to Kimberly Bates and her husband, who are also big time Yankee fans, and we all, we all share in the glory of, of this day. I don't know what'll happen Saturday, we open up against Toronto, but at least for today, there's a lot of joy in Mudville.

Um, but let's, let's jump into things. Let me introduce you to our panel for today. First of all, we have my friend Tracy Reagan from Deploy Hub, the aforementioned Kimberly Bates, Jack Parler, and, and as I said, Mike Vard panel gang members, welcome Mike.

What, you know, as usual, there's a lot of AI in the news. There we go. Well, let's get started with this report from the Yale Budget Lab and the Brookings Institute talking about their analysis of the impact AI has had on the job market.

And they find it's been virtually nil and nobody seems to have lost their job because of ai. And they're basically saying, you know, this is another instance where we have some awesome new IT technology that it gets hyped through the roof and it has negligible impact on productivity. So, Alan, I know you've been following this space for a while, but you know, when I talk to developers and other folks, they'll say AI saves them a few hours a day, but it's really not changing their jobs dramatically, and it's not really having a major impact on GDP.

Hmm. So I'm gonna give a very quantum like, answer to this one. Mike.

I agree and disagree at the same time. You know, I, I did a shimmy says on Friday about ai apocalypse not is the title of it. Here's where I think we are.

And I think this report captures it, and I think your comment is insightful into it as well. I think when you look at adoption of technologies, whether it's ai, the cloud, the internet itself, cellular, whatever the technology is, there are rungs of the ladder o of the technology adoption. I, I think the initial stage is at the individual level, right?

It, you know, do the geeks do those early adopters, do the rank of file geeks find it interesting, helpful, worthwhile to experiment with? Then I think you move from the individual level to the team level. And teams are deceiving.

You could have a team of four to six people, you could have a team of 40 to 60 people. I tend to go with the Spotify model, the two pizza, you know, or the Amazon two pizza or the Spotify I think is 10 or 12. I think that's a good team.

Um, and then beyond that is the enterprise level. I think what, where we are in the AI adoption curve right now is much as an earlier study that we saw last week, and I'm, I'm blanking on who, whose it was again, said that, oh, uh, it was the Google Dora, uh, Google Dora study. 90% of developers are using AI in some form or another, some fashion or another.

I think this study would show a similar thing, but it's being done at the individual grassroots level. And Mike, as you say, they are finding it useful. They are finding it interesting.

They are experimenting with it to make it better, right? But at the team level, I think is where the friction is right now. I think some high performing teams are figuring it out at the enterprise level.

I don't think it's there yet. And I think in spite of CEOs saying otherwise, that's what we're really seeing. So yes, I don't think people are losing their jobs to ai.

I don't think people are being hired. I do think people are not being hired because of ai. Right?

But people aren't losing their jobs. If you're looking at it, the unemployment numbers and what's going on with the guys that are just coming outta college. There's two things that they talked about.

One, when they looked at other, other technology changes, it's a long tail. It has a long tail to coming in, although that the uptick with AI is a little bit faster in terms of some job displacement kind of things. Um, and I think it's, Alan, what you just said was really super insightful, so thank you.

Um, it's a well said kind of position about where this is. The, the other piece of it is that they did talk about the three major industries that are impacted, which is the computer science, you know, computer, mathematical areas, which are absolute environments, which are probably, it's easier for the computer to do because it tends to be more specific as opposed to interpretive or analysis. Um, and so that it can, can do things for them that they're beneficial.

Um, but that's the impact is there. It's just a slow moving piece of it. And that goes back to the piece of saying we are still in the early stages about how AI is impacting our world.

You know, the AI piece of it that I expecting to see is gonna be coming from, you know, some of the robotics, some of the reward, the predictive things. And that takes a long time to train and to assure that the data and the quality that's coming out of that is really good. And until we can have that level of assurance, the enterprises aren't gonna release those kind of capabilities because of what would happen if they have a bad outcome.

It has a huge impact on the company with bad outcomes. It's not just a little bit, it's a big in big impact. And What this article doesn't say, it doesn't talk about the jobs that AI creates.

Mm-hmm. We have a massive amount of money going into investment in AI with new companies being opened and startups blossoming. 2 million jobs out there were really at risk, which is what, like a half a million jobs, right?

That could, could, uh, leave the market. But if we think about who's in the market now, we do have, uh, an aging population in particular in some of these, uh, very, very important roles in, in technology in particular. Um, DevOps, uh, platform engineering.

There's a lot of old folks in the, in those areas, although a lot of young people are starting to go there. So we need, right now, we need AI to pick up some of the pace for us and AI's making our jobs easier and we're better at it. I just haven't seen a lot of people lose their jobs because of ai.

I don't know anybody who has been lay off, laid off because AI took over. I, I don't know, a single person. However, That's not what we're, That could be the case.

That could be the case in customer support. There are some jobs that could be being impacted that this, this particular article doesn't reference Tricia, I think. And also customer support.

You have a huge churn. So customer support, depending upon the, the market that it's in, customer support or customer service, um, phones are, can be up to a hundred percent turnover in any one year, much lower for the higher capability space. So you might not see the unemployment going on as much as you see the lack of hiring.

I think the big risk, and, and Tracy, I'm glad you brought up that sort of the age thing. I think the big risk is the perception right now is that AI is capable enough to take an entry level job or a junior level job, where we don't necessarily need an entry level or junior level job. And right now, while we do have an aging workforce, we also have a very big age discrimination problem in this, in particularly in technology where it's very hard to get hired if you're over 50.

And it's actually such a problem in France that France is running advertisements right now trying to convince employers to hire people over 50. Uh, the French government is, but from a technology perspective, if we don't hire junior people, new college grads and junior people into the industry, they don't become mid-level or senior people. So we run out of the mid-level or senior people, and then we're gonna be in a, in a really sorry shape in three to five, seven years time when we just don't have anybody for these jobs that we need to fill to fill that can't be filled by ai.

And that really scare Me. This is also a channeling indicator, right? Um, because we're talking about usage of AI in the last three years, which is all this copilot kind of stuff, which probably is not nearly as sophisticated as AI agents are gonna be.

So this story is not over yet. It seems to me like the next generation of AI to Alan's point, is gonna be a lot more capable than the first generation. Yeah.

But you know, nothing unusual about this, right? It, this doesn't mean AI is a failure, you know, it doesn't, it doesn't necessarily guarantee its success either. But this is the, the way of things grasshopper, you know, it, it, it needs, it goes individual team enterprise To a, to a certain degree, you're, that is true for those of us who are familiar.

But, you know, once again, here's the IT industry running around making claims about things and credibility goes through the floor because they don't get realized. And you know what? People stop listening to the IT people because they go and they don't.

Yeah. So I, I, uh, it'll be interesting. It's gonna be interesting for the next two, three years on this front as it continues.

Probably longer than that, Than that. I think it's also like the AI has a huge impact on worker productivity. And I think we were probably, if we look back in history, we'd probably see the same things as computers entered the workforce.

And, you know, there's an entire group of people, my parents' agent that were secretaries and uh, uh, administrative assistants that no longer have, there's no career there anymore because most of that work can be done by the individual executive themselves using, uh, you know, email. Nobody has to type out letters anymore, right? Mm-hmm.

So it's not necessarily that AI will eliminate jobs, but it will change the jobs we have. And, and to that point, Jack, and also Tracy also said that similar thing is that shift in changing the jobs you have or changing the industry we have. So I was thinking last night, okay, so we'll go, you know, go back to the Blockbuster to Netflix shift, which is a change of how the entire industry was delivering the, the movies, right?

So it started out with Blockbuster Go and Shop, and then Netflix came on that they were actually shipping you the CD and then became, became the streaming part of it. And so now we would never, you know, yeah, you still have the red boxes that are hanging outside of CVS or something for somebody to check out something, but it completely changed how the industry does. And that takes a long time.

I mean, blockbuster didn't go outta business day one. It took, you know, almost 10 years for that decline to happen all the way through. Fair.

Absolutely. Actually, there is still one blockbuster left I thought, or it recently closed or something. I remember seeing an article.

Um, anyway, let's take a break here. We, we will, I'm, I don't think we're done talking about this subject. We'll, I'm sure be coming back to it, but we're gonna come back and talk about platform engineering in the age of ai.

You're watching Textron Gang Discover Textron Group, the epicenter of tech innovation. We are your go-to for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us.

You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients. Let's revolutionize your tech journey.

Contact us today and tell your story to the world in the most powerful way with Textron Group. Hey folks, we're back. org have a new study talking about how well platform teams are using ai, which may not come as a big surprise to a lot of folks at this point.

But Tracy, let's start with you on this one. Um, I wonder if the rise of AI doesn't force you down the platform engineering path in the first place. 'cause I need some way to centrally manage this stuff at scale.

And are these two things joined at the hip? I think that the platform engineering movement would've come along without ai. Um, centralizing all of these pieces and parts has been something that we've all tried to do for quite some time.

Um, and maybe AI has made it easier to do, but we would've figured it out without ai. I wanna say that, that, that first. Um, but you know, as a follow up to the, our last conversation, I think that there is a, there's something in this article that, um, kind of references what we were hearing in the last conversation.

And that is that there is sort of this gap between, um, these productivity gains that might allow us to lay people off and something that's measurable, uh, that really has improved the way we do work. Um, when you have, you know, the one thing about platform engineers is they like to tinker and they are very careful with their tinkering. And when they start generating code, they're not going to just use it.

They're going to use that as a beginning, a starting point. So it may make them, uh, make it easier for them to do their work. And while there may be new, new, new tools that are coming onto the market, um, that consolidate this information better and does a better job of managing prompts, we are still seeing a lack of trust in what AI is delivering.

So this was discussed in the last, in the last conversation too, when you have, um, when you know when you generate something and you know what's wrong, you may not do it again. So hallucinations, uh, and literally just the skills ga and prompting is probably slowing down the rate of in which platform engineers trust what they generate or what, what AI they're using. And they may continue doing it on their own until they feel more confident in the product that's being delivered.

Uh, I know on, on our side of the house, we we're constantly kind of having jokes about the, you know, the delinquent schoolboy that will just give you any answer because it wants to give you an answer. But if you're a platform engineer, you don't want that. You want something solid and something you can trust.

And there is a big trust issue here. Yeah, I mean, to your point, all this AI stuff is still probabilistic and we're getting a little bit better at making it deterministic, but that takes a lot of skill, effort and time, and I think a lot of folks don't have the patience for that. And I cannot tell you the number of times where, for every one success story that I managed to do with ai, I've probably got 10 where I just threw up my hands and said, you know what?

It's faster if I did it myself. And you know, at platform engineers, DevOps engineers, we are so easy. We, we do that quicker than anybody else.

We are like to tinker, we like to script, we like to sort things out on our own. And, uh, AI sometimes just gives us a bad taste on our mouth because we're like, okay, I just spent all this time working on the script, I could just written it. And much of the scripting that we see coming out of the, that kind of scripting when we're talking about platform engineering scripting, that's different than writing code.

So I wanna keep that, you know, that should be, it's platform is code. We haven't gotten there. It, it's platform is code.

It's different from writing a piece of source code is way different than, you know, setting up a, I don't know, a frame and some widgets, right? It's very different. I, I think it's, look, this, this use case right here is the poster child for what I spoke about in the previous segment platform, engineering platform by, by almost definition is enterprise.

You don't really need a platform. When you got a handful of developers and two DevOps engineers, you start running in DevOps started running into scalability issues when you started bringing on hundreds if not thousands of developers, and you move DevOps from the team level, right? Bubbles of DevOps throughout an enterprise into an enterprise wide DevOps, uh, deployment.

That's when you started running into sort of scalability enterprise issues, which gave rise to the whole platform engineering movement. Okay? So by definition, platform engineering is, is enterprise.

Now that's, and that's where I think the AI piece of this is, is butting heads because it does offer the promise of making our platforms, um, more automated, uh, more intelligent, even if, if that's how Oxy mark. But it's one thing chase for, for a couple platform engineers, couple DevOps folks to, to do some scripts. It's another thing to scalable scalability, making that enterprise wide.

It reminds me, you know what, Mitchell and I were two of the three co-founders of still Secure. Our first product was an intrusion prevention, uh, I IP IPS system, intrusion prevention system. The original product was written in Pearl Scripts, right?

And we, and we quickly realized that, that, that you can't go to market, you know, with a bunch of Pearl Scripts. We, we had to convert it to real code and it didn't take long, it took six months or whatever. But that's where we are.

I think Tracy, that's the exact place where we are from the individuals playing around with some scripting, doing some platform as code to saying, okay, how do I institutionalize that? How do I enterprise that? How do I make that scalable?

It can't just be some pearl script And as long, and if we're trying to push AgTech ai, right? To help us make these decisions, and you have AI hallucinating, the last thing a platform engineer wants is something that's gonna make a mistake. So is there a, it's Not, it's not an option.

Is there an I, and this is a question really, Tracy for you or others. Is there a difference between the approach of how this is used? You know, when I think about, you know, legal issues or sales issues, et cetera, if you have one wrong statement, your credibility is just thrown out the door, you just might as well give up and that kind of stuff, because they're not gonna believe anything else you say.

So that's one thought process that when we look at ai, one wrong thing is like, I can't trust it anymore. I can't trust you. I can't trust what's coming out.

The other side of thinking is that when I look at ai, I know, if I know that it's gonna come out with something wrong, but 90% of it, or 80% of it is right, then maybe I look at it differently. I look at it more like a junior person delivering to me their project that I've asked them to do. And knowing that I have to go through that work and make sure I correct it.

And I don't know how much of a heavy lift that is. It depends upon how good it is. But is it a different thinking about how this is used as opposed to saying, I want this to be a hundred percent right when it comes out.

I think that for most platform engineers, they want it to be right. A hundred percent. And and I think that Mike brought up something very early on that is very important is people need to understand the ca how the tool works and the capability of the tools and ai, what we're talking about here is really large language models, and they are by design non-deterministic.

They're actually explicitly designed to not give you with the same answer, with the same inputs every time. And if you don't understand that, you're in for a world of hurt. If you do understand that, you can, I think as Kimberly said, use an LLM and use an AI as a coding agent to give you a foundation upon which to work, right?

If you just take the output and say, I'm gonna put it into my production environment, never look at it, never trust it and just assume it's right. That's the wrong approach with this particular type of tool. It's not a compiler, right?

A compiler is guaranteed to give you the same thing every time. And we believe compilers to give you to be correct, right? They don't make mistakes.

But an AI is designed to give you that non-determinism. And so if you think about it that way, you can say, okay, I don't necessarily understand how, how, how to do this. Or here's, I can use the AI to generate something quick and dirty from which I can then edit and base it.

I use it as the foundation. That might be a better way to treat the tools. Now, I don't know, Tracy, if the, the, the platform engineers look at it that way.

Probably not. I would rather take a template that somebody's already used, and I know there's no problems. And, you know, now if AI could take my template and substitute all of my variables, that would be great, right?

But it doesn't do that. And that's, that's how, that's how a platform engineer is gonna think. You've already got a template that everybody's agreed on and it works.

Why generate a new one every time? So there's there, there's something about the, this industry is different, this part of the industry that's different. I think it will get better, but we need to think about it in the same way we think about virtuous cycles and DevOps loops and all that other stuff.

'cause I think what will happen is an AI agent will write a piece of code, it will be verbose and full of vulnerabilities, and another AI agent will review that to identify the vulnerabilities and remediate that. And then another AI agent will be functioning as an SRE that will kinda optimize that code for deployment and the production environment based on Tracy's template. We're not there yet, but I think that that's ultimately how that might play out.

Even after all of that, though, there's two factors. I mean, running that number of processes through AI agents is expensive. And two is, I still might not be at a hundred percent, I might be at 95%, but let's be honest, there's a lot of humans out there who are writing code at 70%.

So is that better? I don't know. Fair enough.

Here's the bottom line. We're we, you know, I used to say we were at the beginning of the beginning of the Ai ai story. Maybe, maybe we're moving to the end of the beginning, but there's still more to this story till we reach the end of the end.

And, um, it's gonna be interesting. That's for sure. I have faith that AI will finally be what we want it to be.

It's just not there yet. And I'm gonna say it again, large language models are not domain specific, which causes the hallucinations. In that article, they said 53% of, uh, platform engineers, you know, struggle with hallucinations.

So there you go. AI folks, Hey, give us some clean data. Start using small language models and maybe we'll get better.

You better. Well, I, I think that that is a big key of it, right? Because we, I mean, even things we're looking at here at Tech Trunk, we realize I don't need the whole internet.

I, I just need a corpus of knowledge that we're gonna work off of. But anyway, hey, let's take a break. We're gonna come back and we're gonna, well, we'll still talk about AI in some form or another, but it's, what do they call it?

Physical ai. com is the leading resource for news analysis and education on challenges facing the cybersecurity industry. com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more.

com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more. com to learn more. com.

Home of security bloggers network. Hey folks, we're back and we're talking about robots and security. Jack here has an article up on Security Boulevard that you should all check out talking about what are some of the issues we're gonna encounter, and they are already substantial and some of these robots are even in our homes.

So, Jack, what's going on here? Well, let's start with what we're talking about, which are robots. If anybody is familiar with the, uh, you know, you've seen many of the clips on YouTube about the Boston Dynamics robots that do flips and jumps, or they're one that looks like a dog.

This is essentially the same thing. It's not Boston Robotics. It is a Chinese essentially clone of these companies unitary.

But there are many of these floating around, and these are now very sophisticated, um, not toys. This one is a $16,000 humanoid robot that's about the size of a 10-year-old kid, about four and a half feet tall, 75 pounds, and can do a lot of stuff, dance, whatever you want it to. Uh, the security side of it is, I think a lot of people would look at these things as iot devices, that they're just, you know, that they're just a, a, a dumb network device, but they're not a dumb network device.

They're actually very smart, very complex with a lot of different communications capabilities. So in this case, this robot has, uh, wifi and Bluetooth plus, it has microphones, it has speakers, and it has the, uh, intel real sense cameras, which provide 3D imaging and depth perception as well as LIDAR capabilities. So it's able to do physical mapping of the environment and visual mapping and, uh, all a lot of other things.

The risk now comes in, in that the security on these things, uh, and this was, this information comes from a research article by a bunch of, uh, researchers that was published on our Ziv. And they said it was the most mature from a security perspective that they've ever seen, but it was still very, um, had a lot of vulnerabilities. So some of the vulnerabilities include that the, the Bluetooth encryption was based on a static key that was easily discovered, and that static key is used across their entire family of robots.

So once that static key is known, the entire fleet is com easily compromised. Uh, they use for communication, they use their own version of an encryption rather than using a well-known encryption, uh, algorithm. And that one has already been partially compromised.

Um, right. The robot itself, um, has a fairly complex, uh, fairly capable processor, quad core cortex processor in it, memory, um, uh, onboard storage. And I think the biggest risk here is that it continuously calls home to the, uh, uh, the home, uh, headquarters, which is in China and is continuously, uh, sending telemetry back home.

So from a secure, right? And now it is an, if you thought we were getting away from AI and discussing robots, you're not because this is an AI enabled robot. So there are essentially, sort of two ways you can think of this from a security perspective, or three ways really is one is the robot itself can be compromised, and once it can be compromised, it then can be used as a Trojan horse espionage agent.

So this is a robot that can sit in somebody's office or in a meeting room, and it looks very cute and it's very lifelike, and it's, you know, it's, it's very clean. It's not an industrial looking robot. There are no wires hanging out.

It's completely autonomous and mobile, and it can do audio recording of meetings, it can do video recordings of meetings, it can do document scanning, and it can do facility mapping. So if you weren't in an area where you don't want people to know about that, that's a big risk. The second thing is it has an AI agent built in, or not an agent.

It has an AI environment built in, and it uses AI to do its own tasks, but that also can be compromised and used maliciously. So somebody can actually, um, get into the environment and, uh, use, leverage the AI for offensive measures in your network. So this robot has a lot of AI capabilities built in and has a AI engine itself that it uses for its own, um, operations.

And the researchers demonstrated that you can take over that AI and use it for offensive operations. So they were able to both, uh, exploit, uh, uh, map out, uh, potential exploits of the connections back to the home headquarters, as well as use it to, uh, map out network connections in the local networks and potentially use it for offensive operations. So the big risk here is they have something that is, you know, really cute and simple and seems benign, but actually represents a huge, uh, Trojan horse security issue for an enterprise environment or a military or top secret facility where people don't really understand the capabilities and how it could be used.

So basically when Alan gets, when Alan gets that robo dog, I can hack into it and chase me around the office. Is that what you're saying? But, but that was always, I'm right there with you, Mike.

I mean, but this has always been a, an issue with, with all IOT devices, and I realize our ro our robots are smarter now. And Tracy, shout out to you with your robot shirt there. If you wanna show people your ro robo dragon, uh, there it is.

Um, you know, but here's the thing. You could have always hacked into these robots, whether that robot was in a GM or Toyota car factory or some other high precision factory in China or something, right? You the ability to hack in and make it, in essence, a Trojan horse existed.

I think empowering these robots with AI brings the level of potential harm much higher, much higher. And, and I think there's, there's a another issue with having ai, that these robots are AI empowered. Is it AI on board?

In other words, do they have the horsepower processing power storage to actually run the AI on robot? Yes. This is, that's, that's big risk.

Or Do they fall home? Well, they, they do both, but the AI is running locally on the robot, which means that when it is employed for offensive operations, offensive from the robot's perspective, it is capable of operating autonomously and at a speed that is maybe hard to defend against unless you're using AI to defend against it. I mean, this is is attack of the clones.

Yes. So if I look at a acquisition issue, okay, so I'm gonna take it a different, slightly different way. Um, and I'm reflecting on a, a friend of mine that presented on the implementation of looking at security and IOT in the city of New York, and she had the responsibility for implementing this.

It was a huge effort that they had because when they started looking at IOT in New York, you looked at all the cameras and everything else that was across all the depart divisions. What she found in terms of working through this is that the IOT devices, like a robot is purchased by a department, not by it, or it has this overlay piece of it, responsibility for security making things are secure. Those organizations that are doing that or using however the robot is going to be used do not necessarily have the lens of looking at these for how security is.

So from an organization, as they're looking, as there were at ramping up the use of these devices, there needs to be an organization that has oversight to look at these items to say, is this secure? Is this not secure? And, and unfortunately to be a gatekeeper to these, these, these use how these things are used within the enterprises for that reason.

Because you can't expect like a warehouse operations guy to have the all the thought process that he has, should he or she should have in terms of security. It's a somebody that like you, Jack, that has been, you know, drenched in this space that's got the ability to look at these things to say, okay, so this is, these are all the elements we need to examine. So what I'm hearing, it's, you know, we like to point the finger and say, ai, ai, this is the problem with ai.

It's really not ai though. What we are worried about in this conversation is the data that's being collected, it's data that is our problem. Ai, you know, AI is just another way of using data.

So it's data that these things are collecting that we don't want that data out there, but we are already getting used to doing that. For the last however long we've been doing TikTok and FA Facebook, we have been sharing personal data in ways that we would've never thought of before, and we're becoming more reliant on the data that gets given back to us through these ai, um, tools. So, for example, I'll just say something really stupid.

I did, I was looking for a vet in Albuquerque, I live in, in Santa Fe, so I'm not really familiar with that area. And I'm listening to my Google thinking, Google knows everything about maps, and it took me down a dirt road and I went down that dirt road. Oof, I'd done that.

I, I, I actually happened to be in Napa Valley. I went it, it took me to the vineyard instead of the winery. And, and I got like, I'm on the top of the mountain and then a tree fell and it was not a good, I spent a day up there sitting in my car, And it's because of the data.

Now, in this case, it's even more nefarious though. It's, it's sitting into our personal, it's in our bedroom. Es essentially is what Jack is talking about.

So what's gonna be done with that data when it knows how to map out our house? And these robots are not that far away in Australia. There's one called Abby, I don't know how sophisticated Abby is, but Abby is used for elderly people who are living home alone.

Um, so they have somebody to talk to, they can ask questions to it, uh, they can, it will call people for them. Um, and I'm quite certain that Abby's probably pretty networked in, right? So it's already happening is the data that we should be worried about.

Here's the paradox. What a great thing Abby is. If you, if you have an elderly parent, I who, who's alone to have a companion just to even talk to, let alone to do useful things, you know, I had this conversation I ran in a 5K last weekend, uh, to benefit, it's not wounded warriors, but a similar sort of organization for vets.

How many vets, you know, suicide among vets is a crazy number, right? I forget what the statistic is, but how many vets would just have, would benefit from having someone to talk to someone who maybe can even help them navigate the VA and all of that stuff, but also just an outlet to talk to who kind of understand or can, you know, listen, at the very least, these, there's a reason why we look at these robots. It's not just that they do the nifty Boston Scientific trips.

The dog can, you know, fetch a Frisbee, catch a Frisbee, and do somersaults and all that is because the real promise is there's a lot of lonely people. There's a lot of companionship and things these robots can do beyond just welding, uh, cars, right? Making welds on cars, which is what most robots do in Indus in industry right now.

Um, I I saw a thing that, that again, out of a Chinese factory where they're working on a, a robotic face that has like human-like expression, right? Because, you know, and that gets into all this is Asimov stuff, right? Do you want your robots to look like a human or do you want 'em to look like a robot?

Um, but, And what are you gonna do with the DA when it tells you to do something? Are you gonna drive to the top of the hill or down the dirt road? That is the suspension of our own thought process.

Well, having a data turned against, it's not really turning against, the data's not being turned Against, it's becoming over reliant. Not thinking for over reliant yourself, not thinking for yourself, ab abdicating your, you know, thought process and giving and handing it over to a robot. Yes.

Right, A robot. And Alan, I don't, I'm I'm not in any way arguing against robots. I'm much No, you're Jack saying, right?

I, right, but I'm saying that, you know, I think you could sort of look at this sort of like investing, we on the panel here would be considered sort of mature, well knowledgeable people. There's a whole group of less knowledgeable, less mature organizations and people that wanna deploy these things. And my worry is because of how they look and feel and operate, people will assume that they are benign and not treat 'em as a potential, the potential risk that they are.

Right? It's, and so that it, it, um, it has the potential of erasing, erasing our o our, um, fear factor of it, right? Well, well, there's been several, And that's the issue.

There's been several Hollywood movies on this subject, right? The subject. Yeah.

But it's a broader discussion. It's not just robots. No, it really isn't just robots.

It's, it's, but It's Alexa. But There is personal response. It's Just the next version of Alexa.

So I'm just installed the next version of Alexa, and you're right, Kimberly. It is, uh, it got, there is, There is personal responsibility at work here, though. So if you get a robot and you get a firewall for the robot, if you don't get a firewall for the robot, you're kind of an idiot, Maybe.

Yes. But as we, but I also think There's two sides there. There's what you guys were talk, you know, started going down is the individual use.

Yeah. And how we're using and how, you know, that group of people. We definitely, there's a different there.

And then there's the other piece of it is, is Jack was talking about is how the enterprise implements these things. So I think from an enterprise standpoint, because there's a bigger blast radius, you know, when, when I think about an A here and that kind of stuff, my blast radius is not that big. So you look at the return on investment for the secur, for the guys that wanna break into it, you know, their return on investment is minimal there, but the return, you know, it can't, yeah, it's gonna disrupt and it can hurt and that kind of stuff, but the return on investment on the other side of the house is much bigger, um, depending upon what they're trying to do.

So there's there that risk reward kind of how much do you need to button down versus Not? And then if I can, if I can raise sort of one more risk factor is today these things are connected through the corporate wifi, which gives you at least some semblance of control. I predict that the next version of these things will be 5G enabled.

So there'll be cell phone enabled, and then corporate networking will have no visibility into what's going on in network traffic that's going on. So they will become invisible to it and OT and cybersecurity at that point. Fair Guys, we're over time here.

I gotta pull the plug. I'm sorry. What a great discussion though.

I can't wait till we have a robot here on the Textron gang panel. Um, don't laugh, I've looked into it. All right.

Hey, enjoy your Monday, everyone, as usual. We have a full text drunk TV lineup immediately following. So check that out.

If you're not watching this on Monday on our stream, maybe you're on text Drunk TV or the text drum tv, YouTube channel or OTT channel, wherever you're using it, I hope you've enjoyed, stay tuned for more and we'll be back tomorrow with more gang members and more great stuff. Until then, go Yankees, this Alan Shimel, we're out. Hey everyone, it's Shimmy and welcome to another Shimmy.

Says, uh, sorry for not missing, uh, doing, not doing it on my usual Thursday. We're doing this Friday, but, you know, Jewish holidays and all. Um, anyway, I wanted to talk to you all today about the AI apocalypse or AI apocalypse.

We're hearing more and more about just AI being the root of all evil. It's destroying our jobs, it's destroying the minds of our youth. It's making us lazy.

It's not working like it's supposed to. It, you know, and, and guys, you can't have it both ways. Either it's really good and it is doing all these things, or it's not really good and it's not doing all these things, but to say both.

Well, I guess it could be, um, you know, if we're in the quantum universe, it could be both yes and no. Right? But Quantum Q day's not quite here yet, though.

I have heard about how AI is actually gonna be the catalyst for, for quantum. Um, in any event though, you know, I I wanted to bring my own outlook on this. This is gonna be, I have a birthday coming this month, big birthday for me.

And I've, you know, it's po it's made me reflect, it's caused me to reflect on 35, 40 years in business. And all of the things I've seen come and go in water under the bridge. And I'm not saying any of it was, or most of it, or all of it was anywhere near as disruptive as AI can be.

But I don't think we need to rewrite the rules. We're not in the new paradigm. It's not the new normal.

These are things that, you know, I, I think the AI adoption curve and what AI's going to do is going to follow, at least in a broad level, the, the outline of what we've seen other technologies do, maybe in a bigger way, maybe not, but it's still gonna follow that model. And I, I think we, it's very hard when you're in the forest, right? You see trees, but you don't see the forest.

And, and this is a problem, this is a problem. I think people need to see the forest for what it is. AI is, is helping us, right?

I recent survey we talked about on the gang this week, 90% of developers are using ai. Something like 40% don't trust it. 66% say it, it increases instability, but yet 90% are using it.

So if you have those kinds of numbers, what is that really telling you, right? It's telling you that you, you people are using it. They're seeing value in it.

Where, where I think the issue is, is we haven't been able to translate that value from an individual using it to a team, using it to an enterprise using it. And make no mistake, there's a difference between usage of an individual to a team, to an enterprise, right? At the enterprise level, it's gonna be really scalable.

It's gotta be really locked down. The processes have to be in place. The team is sort of that in between, between the individual and the enterprise.

I'm gonna come back to it, but at the individual level, it's about experimentation. It's about trying it out. It's about trying new things, trying it in different ways and seeing how it works.

It's a really great thing, and I think it will continue to be, um, at the team level, you need it to be flexible enough for several individuals, but you don't need the full scalability, the full process driven type of implementation that you see at the enterprise level. And that, that triumvirate right there, those three different use cases is I think what's driving a lot of the, of the churn and burn a lot of the discussion in the AI market today. I think at a team level, it's succeeding as you would expect it to.

Well, excuse me, I think at the individual level, it's succeeding as you would expect it to. I think teams are figuring it out or trying to figure it out and making some progress. I think at the enterprise level, look, there's always a lag.

I think it's gonna take two, three years for us to really see this at the enterprise level. But in the meantime, you've got CEOs who are laying people off, not hiring as much if you believe the, the press, because AI is going to do it. And, and there's, there is this AI gap, there's this gap between perceived functionality, perceived, uh, you know, abilities versus the reality of it today.

And, and it's gonna make for a rocky, bumpy road. Make no mistake about that. This is gonna be a rocky, bumpy road, and I think we're gonna have to kind of, you know, strap ourselves in as we, we ride over that road.

But for those of you who are consultants working in small teams, individual contributors, I'm every CEO, every exec team I've seen is passing the word down, experiment with it, use it, see what it could do, because we're still kind, stretching the envelope. You are stretching the borders on how this all comes together. So AI apocalypse, not ai, enterprise grade, unfortunately not either, but lots of good use for AI at the team level and in the individual level.

You bet. That's Shimmy says for this week. Have a great one everyone.

We'll see you next week. Says, Hey everyone, welcome back here to Textron tv. I'm really happy to introduce you to another first time guest here with us today.

He's the CEO of Fleet device management, or fleet as they call it. His name is Mike McNeil. He's the CEO.

Hey Mike, welcome to Text on tv. How are you Ben? Hey, Alan.

I'm good. Glad to be here. So Mike, I guess we should start off with, with you, that's a good place to start.

How, how did you get to be CEO at, at, uh, fleet and give us a little bit of your journey. Yeah, so I came from, in 2013, I built this open source framework, um, and it was got real popular, kind of rose with no js, um, got me into Y Combinator and into the world of, of startups. And, uh, along that journey I met this guy, Sid.

Uh, and turns out, you know, this was in 2015 is when I met him. He, uh, I sold him my office furniture and, uh, a few years later, it turns out he's the CEO of GitLab, right? Um, which is now a, you know, multi-billion dollar company.

The dude was about to go IPO, um, and he came to me and was like, Hey, uh, I wanna start a company. Have you ever heard of O Query? Um, and I've, no, the answer was no.

I started looking into it and it looked a lot, kinda like what I had built, uh, when I was working on this open source framework because it was sql, right? It would spit SQL statements at computers and then the computers would tell you like, here's the processes that are running and here's the apps I have installed. And you could join those all together and, um, really solve a lot of problems.

'cause you're not doing something specific for Mac and Windows and Linux anymore. You have kind of a unified language. Um, and I got excited about it.

I was like, man, this has a lot of potential. Um, so in 2020 we started Fleet device management with the, the aim of taking that tech, which my co-founder Zach built, um, back at Facebook, back when it was called Facebook. And, uh, and we really set to work it, okay, what do we need to do to build the best MDM um, out there that's gonna kind of be open source, top to bottom and take care of all the stuff you expect from an MDM MDM, like remote lock and remote wipe and installing apps and controlling profiles on computers and all that stuff.

Got it. Um, so it turns out Sid's actually a friend of mine too, that, and of course you're talking about Sid, Sid ndi, who is the founder, I think he's now chairman of GitLab, not day-to-day, CEO, right? Uh, and then of, of course GitLab has iPod and it is, you know, a multi-billion dollar the pillars in the DevOps space.

But, um, you know, Mike, the, the, the, the device management space, it's something I was sort of tangentially involved with back, you know, before I was doing media stuff. I was a co-founder of startups too, and we, we did a, uh, one of the companies I did was a security company called Still Secure, and we did something called nac, network Access Control. And, and part of that was the device management coming, but out for us, the choke point was when devices came on the network, right?

Then we could make sure that your device sorta had the, you know, the golden config, if you will, where, and you know, obviously different ones. And then we would as then assign devices to specific network access based upon who the user was, what kind of device it was. And, you know, other policies.

So I'm, I'm a little bit, a little bit familiar and, you know, we would also check the device to see what did it have forbidden applications was what ports were open and all kinds of security settings. But that's been kind of the state of the art. So that, I'll tell you the truth, Mike, that was in 2005, Right?

We, we came up with, with what we called safe access, uh, that was the state of the art 2005 to let's say 2007, 2008. Here we are in 2025, what is, what is mobile, or not just mobile device, but what is device management about today? That wasn't the case back then.

I mean the, it's funny 'cause a lot of the same use cases are pretty much exactly the same, right? It is just that the, the infrastructure has kind of changed out from under us. Like the network is still important.

Um, mostly for us nowadays, it comes up for VPN access and wifi access, right? Getting right certificates on the computers, um, to, to enable that and really making an, an employee's first day as perfect as you can, right? If you get it perfect, maybe it'll be 80% of the way there, right?

In, in practice. Um, but I think that, you know, you can't really rely on an employee being in the network at your office. And even if you think about modern offices with open floor plans, like you've got people connecting to wifi all over the place.

You've got people that go out to lunch and bring their laptop with them. You've got people that work from home, um, and, you know, people throw around the, the zero trust idea, right? And no matter where you are in that, on that journey, um, at the end of the day, you've gotta support computers that are not necessarily on your network these days.

But meanwhile, you've gotta do all the same things that you did before with your identity provider, right? Like you still need to have, you know, Val's joining. Um, he doesn't need to get a bunch of apps that are irrelevant for his job.

Um, right? Like, so we either need to install a lightweight set of apps, or we need to go connect with the IDP and install the right apps on the right people's computers, um, out of the box. And obviously you need to make sure they're patched, make sure that they're, the computer has a local HTDP server turned off.

Um, if you have compliance standards, you gotta meet. Um, really the, the simplest way to think about it is you've got a bunch of different use cases that people take these devices and, and run with, right? You've got corporate laptops, you've got, uh, kiosks inside of office buildings, you've got VDIs that some people are using to connect and, and, uh, over the browser and the list kind of goes on and on, you know, bring BYOD, like employee smartphones, you bring to work, you still wanna get to your email on it, um, all the way to production lines and, and all these different devices.

At the end of the day, you need to be able to provision them like set up, uh, you need to be able to control them, like harden them, do governance, implement controls. You need to run audits, you gotta do support as well, right? That's always been a thing.

But these days, with the amount of patching that's going on, it is just a massive amount of, uh, patching you need to do. And there's a bunch of edge cases that pop up. Um, you gotta perform maintenance when it, when the battery life goes bad on a laptop, and, and really most of the time you just refresh it and just send 'em a new one.

Um, and then you gotta recycle it, right? And manage the assets. Um, some companies care more than others about the CapEx, but being able to track, you know, and at least whether it's in ServiceNow or a spreadsheet, um, where all the computers are, so that finance has an idea of that.

That's, that's all part of the picture too. So the bigger you are, the more complicated it gets. No, no doubt about that.

No doubt about that. You know what, I think also one of the bigger changes, Mike, from back then and now is the acceptance of open source, of open standards open. You know, and this is something you've been involved in, it's something quite frankly Sid built GitLab on, right?

You know, I, you know, when we did the NAC thing back in the, in the early two thousands, there were a lot of enterprises who had a no open source policy, right? Because they, they, there was no throat to choke, there was no, uh, you know, you, you could, maybe you could get like, uh, support or training, but you know, they, it was, it was too squishy for them. Where, where now we live in a world where open source is, is dominant, right?

How, how has the dominance of open source sort of played into the whole device management game? There's, there's actually probably give you a few answers there. 'cause it touches almost everything in our worlds now, right?

Like you, uh, especially if you have engineers on your team, like they're installing open source stuff on their computer, it is its own sort of, uh, I don't call it shadow it. 'cause a lot of times it's, it's just expected. It's a part of the job of being a developer, right?

It's out the shadows. Right. It's definitely in the shadows.

Yeah. And then you've got people running, you know, AI tools at this point as well. And, um, you've got, uh, just, I just spoke with someone from, uh, rippling actually, and they have, uh, something like three or four different AI agents they provide every, every, uh, every employee with, and some of those are open source, some of them aren't.

It kind of is starting to matter less, right? Which is the beauty. Um, I think SUD and I both share a vision of a world where a lot of the software we run is open source, um, which doesn't mean it can't be commercial, right?

Um, as much as I like Richard Stallman and, and the Canoe Project, I think it's, you know, it's more nuanced than that. And there's a lot of use cases, namely all commercial software that have, has the potential to be open source and kind of isn't. So when we, when we run Fleet, we really just run it like any other business, but we take everything we do and we put it on in the open.

Um, so whether that's our handbook, whether that's the actual source code for the agent running on your employee computers and on your own personal phone, right? So is it, you know, is it capturing my text messages? Like, well, now you can go look if you want, um, and you can actually see all of the code, um, you can see the server code, you can even see the code for paid features, um, just like, just like GitLab.

Mm-hmm. Um, and I think a lot of it with the, you know, when you have traditional software that's closed, there's a temptation to do things right as you need to squeeze profits. As you get a little bit bigger, um, to l lock things down, make it a little harder to integrate, maybe try to force people onto a higher tier to buy another plugin or skew.

Um, and one thing about open source is it's all there. Like, you can kind of see all of that in the code. Um, you can see all of that in the issues.

And so, uh, it kind of has a cleansing effect. Um, it keeps everybody more well behaved. Um, and, you know, this is, this space.

A lot of these solutions are getting pretty long in the tooth. Um, they don't move terribly fast. A lot of them are owned by private equity, which, you know, has its own goals to extract as much profit as it can, right?

Um, but that makes features slow to add. Uh, and meanwhile, you have some kind of competing incentives, right? You've got Microsoft Intune that obviously wants to prioritize Windows devices, right?

You've got Jamf, which has this long lasting relationship with Apple, um, and, and other, and other tools out there as well. Uh, and I think that just the, the thing that presents itself as an opportunity with open source is if you're an IT engineer in one of these places, you can kinda look yourself now and you can get a much clearer idea of exactly what an open source tool does, um, right outta the gate, right? Without having to even talk to sales.

And that's new for this world. No doubt. No doubt about it.

No doubt about it. Um, yeah, you mentioned Jamf, they look, they did a heck of a job in the Mac market, right? They did.

And build a great community, right? Like, yeah, I always joke that I wanna buy Jamf in 2031. Well, who knows?

Who knows? You know, it's a great company though. It, it really is.

They did it, they did a nice job in Mac. But I mean, one of the one, I think one of the things about Open too is you, you'll never have to worry about being locked into a single platform, right? If it's open, you can take what you've done in Mac and maybe ported it to Windows may not be easy, right?

No one, no one said it's easy, but it, it does help you kind of break those silos down. We, we, and when you're talking about device management, especially in today's world where device could mean laptop, desktop, iPad, phone, watch, glasses, I mean, you know, e everything today is connected, including the refrigerator. So, um, we, we, I think you need that sort of openness, if you will.

And I'm not just talking about open source even, but openness, open standards to kind of cover all those things. Totally. I mean, I imagine a world where, you know, if we have children, we have grandchildren, they're gonna be using new kinds of devices, they're gonna be interacting with robots, whether they're humanoid or not.

Like it's becoming inevitable, right? Physical Ai, physical ai, Mike. Oh, okay.

That's the new name. That's the new name for the robot. They're physical ai.

Uh, but yeah, no, and sooner than you think, forget your children and grandchildren, you might have one by 2031. It's looking that way. Yeah.

And, and you gotta wonder like what is running on that thing, right? And in that world, how important is it to be able to remote lock one of these things? Um, and more importantly, to, to see the code that's supposed to remote lock it and make sure it actually works.

It's crazy. It's crazy. Hey, man, we're running low on time, but, um, fleet de Fleet device management, how do people get to the website?

How do they reach out? How do they interact? Yeah, so we're on GitHub if you're interested in contributing.

And then, um, we recently raised a series B, uh, and we have a, you know, we have all the things you'd expect. com. Um, you can read our handbook.

We have our open positions listed there too. Um, and we're hiring. Very cool.

Hey, Mike, thanks for coming on Text on tv. Say hello to Sid for me. I haven't spoken to Sidd in a couple years now.

I bet. But tell him I said hi. And best of luck with Fleet Device Management.

Thanks, Alan. Thanks for having me on. Alrighty, Mike McNeil, fleet Fleet, fleet Device Management here on Text R tv.

We're gonna take a break. We'll be back. Our online interactions include audio, video, and sensor data, but most AI applications are still focused on text.

This episode of Utilizing Tech considers how we can integrate multimodal data with Agentic applications. With our conversation with Vka Gupta, founder and CEO of Aperture Data, Frederick Van Hern and myself, Steven Foskett, welcome to Utilizing Tech, the podcast about emerging technology from Tech Field Day, part of the Futurum Group. This brand new season focuses on practical applications of AgTech, ai, and related innovations in artificial intelligence.

I'm your host, Stephen Foskett, organizer of the Tech Field Day event series, and host of utilizing Tech Now for nine Seasons. Joining me this week as my co-host is, uh, Frederick Van Hern, who's been present for a lot of those seasons. Welcome Frederick to the show.

Yeah, thank you. Once again. I'm, I'm, I'm here as a co-host.

So my name is Frederick Van Hern, I'm the founder and CTO of ens, which is a HPC and an AI consulting and services company. And Frederick and I have been talking about practical applications for AI for a long time. But one of the things that always sticks in my craw, I'm not sure what a craw is, but something sticks there, is that when people talk about ai, they too often focus only on text.

Basically, it's chatbot or bust. And that's great. And in fact, there's a lot that you can do with text, but text isn't the whole world.

Uh, and Frederick, I mean, your background in ai, you, you, you didn't start with text. No, definitely not. You know, my, as you know, my background is in speech, and, and that's the language we use to communicate with people, but we have to understand that people are more visual than learning from text.

So a multimodal approach to ai, it's definitely something we're looking forward in this agentic AI world. Absolutely. And I think that, um, you know, just regular people who are gonna, uh, be sort of wondering why are we, why are we always talking about, you know, documents and books and web pages and stuff?

Why aren't we talking about literally everything we interact with today, which is, um, audio, which is images, which is video documents and so on. And that is why, uh, we have an exciting guest to kick this season off. Uh, today we have, uh, Vish Gupta, uh, founder and CEO of Aperture Data, who is somebody that I spoke with, uh, earlier this year, and they are really focused on multimodal data.

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

Um, happy to. So yeah, as you mentioned, uh, right now I'm co-founder and CEO of Aperture Data. Uh, prior to that, I was at Intel Labs for over seven years, which is where we started working on this problem.

And, uh, you know, just take looking at it, um, from a researcher standpoint, and now through the journey at Aperture Data, it's been more like product and user standpoint and business' standpoint. Um, for my background, um, I have a PhD in computer science from Georgia Tech and a master's from Carnegie Mellon. And I got my undergrad in computer science from Biani in India.

So it's been, uh, you know, one part of life where it's a lot of, you know, computer science, deep research, working, really like, you know, uh, on underlying, um, uh, systems hypervisors. We were one of the first teams to virtualize Nvidia GPUs back when it wasn't even so, uh, popular in, uh, data centers, uh, to, to be, to be offered in cloud environments. Um, and then now this has been a completely different part of the journey.

You know, I sometimes, um, miss coding too, uh, but it's a lot more about people than it used to be before. And most importantly, it's a lot about very exciting use cases and applications that have emerged in this last decade. Just as you know, we have witnessed the progress of machine learning from, just from the very basic, like, you know, the very first image, that thing that make it like, oh, now we can actually automatically understand what's in an image to now AI agents trying to like, you know, order our plane tickets for us to plan this perfect vacation.

Well, it does seem though that, um, the progress of ai, you, you're right, that it started, as Frederick said, it started with, um, with speech processing, uh, for the most part, uh, it, it, a lot of the applications early in the utilizing journey. Back when we started this podcast, we talked about, uh, processing images and detecting objects and, and, and processing video and sound and all sorts of other data sources too, um, you know, bio, um, mechanic information from sensors and so on. But it seems like the prevalence of large language models has just crushed any kind of discussion of anything that's not text.

Are, are you seeing that as well? Um, I think a lot of the practical applications, you know, the, the approach I end up seeing for people a lot of the times is, well, they're asked to go use ai. So this is like, you know, you kind of have to, you know, we work with a lot of, um, you know, medium to large enterprises, some startups, and we see this very often.

There is always this like, how can we use ai if, if they're doing it right, it would be more from the perspective of, there is this business problem, can AI help us here and then watch the process? But then sometimes it's like, we gotta get on the AI bandwagon. There's a lot of funding for it, right?

So there is like a spectrum of people. Uh, but the common thing is like, okay, look, uh, especially when you think about larger companies, data is very siloed, right? Especially if they're collecting, if they're going beyond simple tabular data, going beyond text, that means sometimes it'll be like images, videos, audio, they get organized in places that most of the company doesn't know the, like few people in some team that will have access to it.

So just the process of bringing things together and then making sure that the models are up to par, and then can you actually take all of that and combine that into, um, a model that can give you the right answers? That's a very, um, intensive process and which is what makes it like, okay, well let's prove the value with text first, uh, and then we'll see. Right?

Uh, unfortunately though, the problem is in some cases, text is sufficient. A lot of the times people are like, you know, let's say if they are just, uh, uh, there is an example where you have a lot of PDFs, and granted it took some while to start parsing PDFs to the level of, you know, actually understanding tables and images within PDFs too. It's not, it looks simple, but it's not always, but like, you know, if you're trying to understand a lot of reports that you do internally and enable a rack chat bot, that's the kind of application that you can pretty easily start off with.

Um, um, and, and, and, you know, so you start, if, if you start seeing the ROI good, but if it's the sort of application where a lot of information was stuck in these other data types and you started just with text, it is quite possible that people arrive at the wrong conclusion that AI doesn't work. And I've seen that through a lot of, uh, you know, even before we got to the whole rag and egen stories that we are like Gentech world that we are in today, even when people were just trying to do like, you know, let's say e-commerce personalized recommendation, people wanted to use visual similarity search to recommend products that look like this. Because, you know, we are very visual people, like you said, uh, we're, uh, we look at something and that's what attracts us, not the text description of the product, right?

Um, and a lot of the times teams would start building it, but they didn't have tools to be able to query and see what their image data sets look like. So they would just knowledge not do the, you know, just based on your friend bought this, and so you should buy this sort of stuff. Um, and so the conclusion used to be, well, AI is not helping us.

Uh, and, and so that's why I kind of warned in terms of like, you know, it's a text is a good start, but there are a lot of cases where you got to bring in the other signals and it looks very daunting because the tooling also has been pretty broken. Um, but that's kind of why we are here. Right.

So you talked a little bit about, uh, multimodal. I mean, it's, it's already difficult enough to build models just for text or audio or video, um, let alone the different data types, right? Video notoriously are much bigger, binary, very difficult to analyze.

Uh, text is easy to read, but much smaller than video. So how do you deal with all those different data types and those different models into, into one final model, so to speak? Um, I think I, I mean, I couldn't speak, so the way I look at multimodal data and what it would take to unlock everything that it offers, I look, uh, at it at, in, in three, three sections, so to speak.

One is the model, what you're referring to. And you know, there are a lot of vision language models now that are doing really well. I mean, I'm recently reading all about like, um, generative on the SOA side, but also in interpreting like the, uh, if you look at some of the Gemini output and stuff like that, they can really do a very good job understanding what's happening in image video.

The second aspect is processing. Well, there is no lack of processing today, I might say. Um, I mean, Nvidia really has changed the name of that game.

Uh, and there's a lot of other, uh, inference provider providers and, and, uh, some more niche companies, uh, coming up in that space. Uh, and then the third aspect is data management. And I feel like this is where the biggest gap exists.

If, if you look through the evolution of machine learning, you know, we used to see all these papers where, oh, you know, we are now able to detect like this tiniest bit of, uh, like a dog face in the image. And then you went and deployed it in like a medical imaging sort of scenario. And I literally have an example I ran where the brain lobe, uh, like the brain scan was classified as a telephone lobe.

Um, so, you know, so those sort of things, you, it, they happen because like data has always been the differentiator. The better data you train with, the more representative data you give to your model, the better the outcome is. Um, and it remains true to this day, but unfortunately, the solutions are still not there yet because changing data systems is a very involved and very complicated process.

Um, or it, it doesn't have to be, but that's how we've always seen it. Like, you know, I mean, there's so much, and like people think so much in terms of SQL and relational tables. Sometimes I run into people, it's like they won't use anything that doesn't support sql, but that's not the right approach.

The thing is, what solves your problem if, if AI needs to see all of the data, you need a data foundation that allows you to put all of the data, make it searchable, and make it easy to navigate. That has always been our driving principle behind this, because if you wanna go, um, and you know, earlier we were talking about going from shallow intelligence to deep intelligence. Um, we are, uh, you know, like I was saying, the models really are very advanced now, the processing power, you know, it's growing constantly and, and, and accomplishing a lot.

If we solve the data problem, if we build the right foundation for data layers instead of still cobbling together a bunch of tools that make you inefficient, that, um, create inconsistencies that don't let you scale as much as you can, you're still never gonna fully go into deep intelligence territory. Yeah, it's, it's so true. And, and what I'm seeing unfortunately is that a lot of ag agentic applications are still focused on structured textual data.

In other words, if we're gonna have this process pass data onto this next process, um, in many cases what it's doing is it's devolving, uh, you know, let's say visual data into a, in, in some cases, a large set of texts that describes that visual data in either, uh, freeform text or in, you know, structured, um, data and then passes that to the next thing. Um, is it possible for agents to pass the actual image or some abstraction of that image between, uh, agents in these systems? It, it absolutely is possible, right?

So, um, as we were building, so, you know, as you know, the product that, uh, that my company, um, offers is called adb. It's this unique vector graph hybrid database that we have purpose built for multimodal ai. So there was always this aspect of, you know, if someone says, I wanna find images similar to this image, there's of course the vector search angle, so you need embedding generation and things like that, but how are people gonna give you the image?

Because they would have to give you the image to put in the embeddings, and then you can do vector search, right? They would have to, um, and let's say you were going even further, you wanted to find video clips that had, let's say, kids playing in it. You would have to be able to understand components of the videos, um, you know, generate embeddings from those and be able to search through them.

Um, but it's not just the vector search part, because what ends up happening is, let's say you want clips where, um, uh, kids are playing in the video, right? Um, maybe the entire video is 30 minute long and there is only like a two minute section in which the kids are playing. A true multimodal AI database should allow you to search, decode the video, and go into those two minute part and just transfer that.

Now, why is that important? Because I'm gonna imagine video, like you mentioned earlier, videos are really large. Are you gonna be transferring the whole, like, you know, 30 minute long video between different components that need to operate on it?

Or do you wanna just take the two minute sections that are relevant and pass those along? So that's like, that there is, there is this whole efficiency angle to it, and, you know, not having to wait for hours for something to happen that can only be enabled when you introduce true multimodal understanding in your database. And that's, that's kind of what we did because, um, you can literally say, I want all the video clips in which a person was smoking or not smoking, uh, and I want them returned to me in thumbnail size.

This is one query to aperture db. It does the decoding, it ticks out the parts that are interesting and it, you know, bundles up the clips and sends them to the, uh, to the next stage. Um, and you are not duplicating any of this information because remember, you gotta think about scale.

We are gonna operate, we we're operating on petabytes or maybe even zetabytes of data, right? Um, and when we represent videos in our database, that is the original video file, but then we very smartly use the graph structure that we have to represent all the regions of interest in it. It can be interesting frames, it can be interesting clips in it.

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

Um, so to your, to your original question in terms of why can't you transfer some of these other data to, can you transfer these other data types? I think one is the protocol that allows you to define the stuff, and I think there is still some room to improve. Like we had to, um, come up with a different query language to support all of this.

Uh, we actively chose not to implement into like A SQL or a cipher based query language because they were too restrictive in what we were trying to do. Now we have built plugins to make them compatible because a lot of other tooling lives in that world, but, uh, we started out without hindering ourselves, and, uh, we had to introduce the exchange, like, you know, okay, this is how you are gonna give us blobs of various types. This is how we are gonna DeMar one block from the other, and the metadata we return is gonna tell you what the rest of it means and things like that.

So, um, we are able to do it. Um, and, you know, we, um, work with PDFs, audio images, videos, and, um, it, it, it works great. And of course, in the backend, uh, we've introduced the whole performance and scale and, you know, understanding that these data types are different.

You have, you know, more parallelism requirements, there is less dependency among, you know, individual parts of the data. So there's all that stuff that goes into the architecture to make it high performance and efficient. And then there is in the protocol to, to enable, like, to define that language.

So it's, it's very much possible to do it, Right. Yeah, data movement has always been a problem, and as, as, as people collect more data, I think the data movement by itself will get worse and worse. So how do, how do people interact with your platform?

Do you integrate with frameworks or orchestrators, or how do, how do people use and consume the platform? Yeah, I mean, you know, we started out with a database. No one wants to really think about a database.

It needs to be hidden behind stuff. Um, and so yeah, we have, um, uh, you know, originally when we started, because we were looking at a lot of, um, training and inference sort of use cases, we integrated with py, TensorFlow, vertex ai, these sort of, uh, frameworks. Um, then we, you know, with the RAG in the rag world, basically we introduced L chain LAMA index integrations, and now we are looking, looking into agent memory, uh, frameworks to integrate with them.

Um, we also do, you know, so Aper TB has grown from just being a database into this entire platform where, um, you cannot just manage and search the data. But, you know, we have in introduced workflows to make it easy to upload data to, uh, generate embeddings to extract information. I mean, we have a workflow that, you know, you give a URL and outcomes, a rack chat bot, uh, you don't have to worry about segmentation mo like, you know, embedding models and things like that.

Um, so, and, and we have these various, like, you know, MCP server plugins, SQL server plugins, so that has grown, uh, now into a platform so people can interact with ADB directly on our, uh, cloud platform. They can use Community Edition, we can do a VPC deployment. Um, but we also have our own ui.

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

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

So, um, there are different ways that you can go about it. Like, you know, a lot of the agents behind the scenes when they want to interact with data, they basically might, you know, just do vector search queries and then, you know, implement their own LLM like feed, like do the semantic search and feed things to the LLMs and then generate the responses. Um, so that's like very fundamental way, which, you know, you can just use the vector search support we have.

You can enhance that with graph rag, sort of, you know, like rag improvements to start including the knowledge you've contained in the graph. But where we are seeing this go is essentially introducing this memory interface, uh, because if you look at, you know, the memory frameworks, there are, there's the component that actually takes, um, user log user questions and extracts preferences and, and, you know, relevant personas and stuff. But underneath it ends up storing this in either vector databases or a combination of vector and graph database or just simple text logs.

And aper DB is perfect for storing all of that stuff. I mean, the throughput and latency we offer in terms of updates and queries, it's phenomenal. Um, and so it makes a really, you know, good foundation.

And so now the, the thing we are working on now is like, okay, what's that memory layer, right? Like, you know, we start by integrate, we, we will basically integrate with some of the frameworks already out there. Um, so the agents can really like, make use of the memory and scale, um, through what Aperture DB offers.

So I love this talk of, you know, moving beyond text. I mean, that's the, the, the premise here at the beginning. Um, but, um, I wonder if you could help us with some infras or examples or some ideas about moving beyond video too, because of course, multimodal data, it doesn't just mean video, it means all sorts of data types that are, you know, all varied.

So what other data types beyond audio, video, video and obviously text are, uh, people looking at with agent applications, and what are some of the use cases for that? Um, well, you know, I mean, uh, Stephen, as we talked about, a lot of people are still on text. We are not even at the audio video stage yet, but, you know, so there is a lot going on with voice.

So there's a lot of audio information. I think people have realized there's a lot they can even like, you know, even before getting to voice and, and videos, um, there's a lot going on with PDF because they, I mean, you know, if you, um, we, we, we all create so many reports, right? And we like to put tables to summarize, we put charts there, um, we have these pie charts and you know, sometimes we put pictures to show, uh, the way, like, you know, how our architecture looks.

So, so there is a lot that goes in and parsing PDFs in itself is, uh, and, and extracting information to start making sense and, you know, at what bound, um, parts do you, uh, how do you segment it and things like that. All of that stuff involves a lot of, um, work. So I've been seeing, um, like people who have managed to go beyond text, a lot of the times it's like, uh, actually this is the kind of progression.

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

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

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

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

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

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

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

Like, DICOM is like a series of, uh, p and g files, the usual image format. So we would slice it up because DICOM itself contain too much, much. So there is like, you know, um, but that's when you become very domain specific.

Yeah, I'm glad that you brought up, uh, medical, because I think that that's definitely an area where we're gonna see a lot of development in this, but also as, as you talked about a lot of geographical data, um, I was talking to somebody who's working on drone technology and they are working with everything from, uh, you know, GPS data streams to topographic information like you mentioned to, uh, you know, real time feeds, uh, from sensors, and all of these things have to be integrated and localized and plotted together. It was a really interesting conversation a little bit beyond me, but, um, but I could understand the challenges because there, you know, it's not just video, it's not just maps, it's not just text, it's all of these things as well as lidar and radar and, you know, cameras and, and, and all of this had to be integrated. So I, I think that increasingly that's what, what the challenge is gonna be is how do we integrate all of this data in a way that a, um, an AI agent can understand and act on without just overwhelming it with data?

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

So keep an eye out for that. So thank you so much for joining us and listening to this episode of Utilizing Tech. Uh, you'll find this podcast in your favorite podcast application, just search for utilizing tech.

You'll also find us on YouTube. If you enjoyed this discussion, we'd love to hear from you. Please give us a rating.

Please give us a review. Uh, please subscribe. Uh, this podcast is brought to you by Tech Field Day, which is part of the Futurum Group.

com, or connect with us on X Twitter, uh, blue sky, mastodon, or, uh, yeah, LinkedIn. Uh, you can look for utilizing tech. Thanks for listening, and we will see you next week.

Thanks. Every company lives in fear of a ransomware attack, whether they've suffered one or not. And this is even more critical in the era of ai.

This episode of the Tech Field Day podcast looks forward to Commvault Shift in November with a discussion of the importance of data protection to AI applications with Tim Zaka from Commvault and Frequent Field Day delegate, Gina Rosenthal. Listen in and learn about the connection between data, data protection and ai. Welcome to the Tech Field Day podcast, where we bring together a group of IT technical experts to discuss a single idea about key concepts in the industry.

This podcast features a variety of perspectives from members of the Tech Field Day delegate community, and is often recorded in association with one of our events. This episode in particular is recorded in association with Commvault Shift, which we will be attending in just a few weeks. Tech Field Day is part of the Futurum Group, and this podcast is published on our sister company site Techstrong tv.

This episode we're looking forward to Commvault Shift, which is November 19th, and we're talking about the importance of data protection to ai. Yes, ai, it's 2025, everything is about ai, but we've found that there's an interesting connection between AI strategies and data protection strategies. Before we get started with that conversation though, let's meet who's on the panel today.

Hi, I'm Gina Rosenthal. I am a fractional product marketing manager. I help lots of companies including data protection companies with their, um, product marketing.

And I'm Tim Zaka. I'm the Vice President of Portfolio Marketing here at Commvault. And glad to be here with you today, Steven.

Thanks for having me. Yeah, it's great to have you. Uh, as you mentioned, I'm Steven.

Uh, I am the, uh, halftime host of the Tech Field Day podcast, and, uh, have been running tech field day events for 15 years. Over that time, I've actually attended quite a few Commvault, uh, events, and we've had Commvault join us as well at our tech field day events. And one of the things that comes across whenever we talk to folks from Commvault is the importance of data protection.

I guess surprise, you know, I mean that's, you know, what you do, but it's also something I think that is really, uh, in your hearts because you're a company that spends your time talking to companies who either fear data loss or have suffered data loss, and you're trying to help them avoid that catastrophe now that it's 2025 and AI is on everyone's lips and everyone is trying to figure out how to roll out a successful AI strategy. Tim, I wonder if you can start off by just sort of drawing the connection between AI applications and data protection needs. I actually think you just nailed a, a big part of that connection, Steven, is you had said something along the lines of e everyone, you know, focused on kinda ransomware, staying resilient, uh, kind of frankly like the hygiene of what great looks like to just maintain a continuous business and fight through an attack.

Now, while organizations are doing that and putting these best practices in place, they still most still have a long way to go. And you have this meteor of AI coming crashing into that set of projects, practices, and initiatives. So I think one big connection point is as organizations are trying to improve their resilience, their recovery practices to fight through just kind of maybe good old, uh, cyber crimes now, it's just gotten that much more daunting as AI stacks look different, as AI data is distributed everywhere, as there's questions of accountability and ownership and now add just attacks that could, uh, in the threat surface that's widened and different through ai.

And I think that's the big connection point because the so what across all of it, it doesn't matter if it's a, you know, maybe a more traditional looking attack or something that's come through ai, you know, poisoning or even just frankly, an outage due to the newness of it, the business needs to be back up and running. And so having a resilient, strong resilience practice is, I think that the connection point between the two. I love that word resiliency 'cause I think that's really true.

A a couple of things that really stuck with me is it's definitely, you know, AI introduces all of this new ways to attack organizations. I think one of the most interesting ones is that the people sending the ransomware out are using AI to create their, their campaigns. So one of the things we've lost is the, the as just users, individuals that, you know, on the, on taking all the training and looking out for the phishing, you've lost the ability of saying, oh, that's definitely not proper English, that I don't think this is really from that person.

You've now got the ability, the, the, the people have the ability to, to train on an individual's voice and how they write and how they talk and send you a phishing email that looks super, super, um, real hard to explain. And then I love that point about resilience, because you're gonna make mistakes because this is all new to everybody. So you're gonna lose data, you're gonna dump something you shouldn't, and you need to be able to roll back.

So it's, it's really important to think about that. Yeah, and I think you, you mentioned a good point, Gina, which is, it's, I think maybe it's just like we as technologists, it's easy to geek out on the more sophisticated attacks or, or maybe newer ones, you know, prompt injection or just adversarial attacks where maybe it's, uh, where the bad guys are using AI to do, you know, polymorphic attacks where the signature is changing and it's hard to detect. But I think like just making spam that much or, and or phishing rather, you know, that much better.

Um, it's super dangerous. And, and I think that, you know, when you can have these attacks that are just actually logins, not, not these really like well-crafted, uh, uh, you know, initial kind of entry points, um, yeah, that, that's, that, I think that's really dangerous. Yeah, we've been seeing that here, um, where the, i, I don't wanna say sophistication because that's really not the right word, 'cause it's not any more sophisticated than it used to be.

It's just, as Gina said, it's more convincing. And so we're starting to get, uh, well, I mean, we, we're all businesses, we all get ransomware, uh, attacks or at least openings, uh, you know, phishing and, um, social engineering openings all the time. But I've noticed that they really have changed.

Uh, they used to be pretty clunky. They used to have poor English, poor integration, you know, sometimes it was like, dear, you know, last name, you know, that kind of thing. Now it's really not like that, and it's pretty obvious to me that, that it's generative ai, especially with, you know, text chatbots that's enabling, uh, customization and personalization at scale, at least to some extent.

Now, obviously as Gina points out, sometimes they're still a little nonsensical, but, um, it takes a human to recognize that. And I'm concerned that these, um, you know, the AI technology on the one hand, it opens, uh, doors to creating more, uh, credible attacks. But on the other hand, of course, we also have to think about the ways that agentic AI applications are opening the doors for those attacks to be more successful.

So maybe we can get to that second point a little bit later. But I mean, first, um, you know, has the prevalence of ransomware increased or does it just seem like it has 'cause it, it seems like we're hearing about it constantly and is it the fault of AI that these things are, are more credible and that's what makes people, uh, uh, more susceptible to them? I mean, I, I think at least what I see as I, as I talk with, with our customers around the world, I think I, I've seen, I would say most data at least suggests yes, ransomware is becoming more prevalent.

I've seen enough to say, well, you know, it's about the same or even, you know, a slight decline. But I think the punchline is, um, that it's gotten that much more effective and, and, and common and I think dangerous. I think it's, and, and this, it's talking to the two of you, this is cliche to say like, it's become such a massive business that I don't really, I think it's a distraction to think about, like, has it gotten, you know, more prevalent or the same or less like it's a, or a just a staggering, uh, danger and, and a costly one.

I mean, it, it, you know, it impacts business from a cost perspective. You know, we've all read, uh, you know, or, or even some of us like work directly with, you know, healthcare organizations where, you know, it's patients that, that they're, they're working with it's government institutions trying to serve their citizens that are impacted by this sort of thing. So it's, it's, it's like a massive business.

And I think to me, that's the, the biggest one. So regardless, like what the trend is, it's, there's no future where it's going away or, or, you know, diminishing to any kind of significant degree. And I think, Steven, you're right, it's just, it's, it's getting smarter, I think the iterations with which, um, kinda adversarial, uh, or, or just kind of either hackers or usually it's, you know, organizations can just put out a text like it's a business.

Um, I think that's the part that is the most just daunting and challenging. It's just moving so fast. And I think AI allows them to do it that much more adeptly.

And I think you can have people who are, you know, even organizations that are pretty good all of a sudden get really great and they tend to be able to move faster anyway. And so to now have a set of tools that allows them to, um, you know, often outpace the people that they're trying to, um, get leverage against and, and hold for ransom. I I think that's the daunting, the, the most daunting thing.

I think that's the way to look at it too. This is not like the hackers and the hoodies, like the, to crazy way they display attackers it's businesses, it's, it's a business and it's nation states doing it as well. So the, the, the money that they can't get other ways, they're able to get it from ransom.

So they're definitely using every tool they can to number one, build the tools. And they're using, they're using generative AI to write the code, and they're using, they're using it to market. They're using all the tools we use to make sure people hear about the products.

They're like, we're gonna make our product really awesome and we're gonna make sure you hear about it and use it whether you like it or not. So it, it's not, you know, like script kitties, like we used to always think this is literally well organized businesses that want to be well funded, and it's really easy for them to get funded because of the nature of ransomware if they don't have protection. Yeah, that's, that's a really interesting point.

Like Tim said, I mean, this is, these, these are basically businesses now, uh, they're illegal businesses, but they're businesses. And so by, uh, whether they're funded by nation states or funded by ransomware payments, they have a huge amount of power, huge amount of money. Um, they've got this incredible technology in their hands.

I mean, Jeannie, you, you know, as you say, a lot of them are using, uh, generative AI to write code as well as to, to write text. Um, you know, we have a whole new world now of, uh, very, very convincing generated video and audio. Uh, I know that we've all heard the stories of, you know, ransomware attacks that, uh, appear to be legitimate employees on a Zoom call and things like that.

I mean, this stuff is just wild. And, and I think that basically, you know, we have to move on from this idea that somehow we're gonna squash this before it happens, and we have to move to the idea of what do we do when it happens? Because I just don't think we can stuff, this genie into the bottle, There's too much money at stake.

It's not going back. You have to protect yourself and have a way to recover. You have to, it's gonna happen.

Yeah. And I think the whole notion of Yeah, like stuffing the genie in the bottle. I, I think the thing that, kind of back to your, one of your original questions, Steven, is just, you know, that the connection of AI and the impact that it's having on, on resilience is not only, like, we can't stuff the genie in the back, in the bottle, but the, the pace of change that AI is bringing is so dramatic.

At least it's the fastest and most impactful thing I've seen in, in my tech career. I mean, there's, there are analogs to I think the cloud world. I, I see things like, you know, remember in the early days before, like the, um, kinda shared responsibility model was really something that was wholeheartedly, you know, articulated, understood practice, and it's kind of who owns which pieces.

I mean, I think it's a similar thing here in this, this AI world, and it's, it's moving so fast that is org. Every organization that I've talked to recently is doing something there, and many of 'em are doing real stuff, but they're like, Hey, look, I was talking to a CISO the other day. He is like, we use Zendesk for these sorts of things.

And, you know, we understand how to make that resilient, but they're using, you know, what about the supply chain of what they're using? Where does our accountability go? How far does it reach into the LLMs and the, the way that they're, they're training stuff and where does that line stop?

And I think there's a whole set of questions around those kinds of practices that are being entertained now, don't have clear answers, and the pace is moving so fast that it's, you know, I think, you know, people are, are trying to keep up, but it, I think that's the part that's, it's not, it's like the genius outta the bottle and watch out because, you know, it's coming really quickly at you. I think that's a really interesting point too, because, um, PE businesses want to move ahead and it's obvious that AI is the next round of, of innovation for computer science. So they want to move ahead, but everything's coming so fast.

Everything all at once, right? How do you, can you, you still have to maintain that data center hygiene and keep everything protected and, and do that side of the business and, and how do you do that protection and, and how do you know where to go? Where do you put your funds to actually do more than what you've been doing and, and, and use AI to get you to the next level.

Yeah. And I think to me, that that's the biggest tension that I see with the customers that, that I'm working with in that k to your point, Gina, that some of the, the practices that they're putting in place to protect their data, um, even basic ones, things like, you know, air gapping with immutable and indelible copies of your, of your critical data. Things like, um, some sort of practices, processes, technology for identifying clean points to, you know, to recover to, you know, Stephen, as we were kind of prepping for this, do you know, just talking about like, Hey, how far do you, you roll back certain sets of data?

Like, so having a process around what that looks like. Things like, are you class, you know, discovering and classifying your data? You can't do rag if you don't what your, what your data is where, you know, like, you know, and, and, um, and, and I think basics like that are still being implemented, let alone more advanced technologies or, or capabilities like, Hey, I wanna use the cloud to burst into isolated clean rooms for forensics or something like that.

Just the basics are still being put in place, at least broadly. And, and now just this rapid progression of ai, it, the, the, to me, the heartening thing though is the organizations that I am, I am, that I'm, that I've been working with is just, they're, um, kind of delightfully a bit more progressive than I would've thought on, you know, who's involved or how many appli, you know, kind of AI driven applications they have in production. And even when they're, they're moving forward with almost like this blind urgency, uh, and kind of prioritizing learning and, and growth over kind of the guardrails.

They, uh, I feel like at least some of the proper guardrails are starting to get it put in place. And the ones that aren't, they know what they are. Like Gina, to your point, they're like, yeah, we know that the right hygiene looks like this for our established workloads and, you know, but we're three months out on that for this project, or six months out.

So, uh, maybe it's the eternal optimist in me, but I, I at least feel like people know what they should be doing soon for those workloads. Yeah. It's not all doom and gloom too, that's the thing.

So, I mean, AI is also a very powerful tool for data classification and for mm-hmm. You know, detecting, uh, attacks as well. And I think that that's all something I th that, that gets lost sometimes.

'cause we are very scared of ransomware attacks. I mean, I, I am, you know, as a business person, uh, I am as a professional worried about what the impact of these things are. But there is some reason to be optimistic that companies are, you know, kind of getting their act together in terms of data classification, in terms of implementing, uh, good strategies around data protection and that that the tools are advancing as well.

I mean, one of the coolest things out there, and, and it's funny because it's not really a new idea, but it found new impetus, is this idea that you can have, uh, almost real time data protection. And you have that kind of virtual dial where you can sort of go back and say, well, the ransom attack happened here, so I want my data set to go back to right before that. And there again, I don't want it to sound too much like an ad for Commvault products 'cause I know that you guys have stuff that does that.

But you know, that's a pretty powerful tool and something that has finally found some currency in here. And the same is true as you point out of tools that allow you to, um, organize and classify and categorize data to replicate data to different locations, to burst between locations. A lot of this has been something that we in the industry have been building to for a long time, but here it is, and it's actually important, more important than ever when you're talking about these data-driven applications.

Yeah, I think you bring up like a, one of the kind of current hot topics, I'd say in a lot of discussions with our customers at least, that are, that kind of are, uh, running their applications regardless if they're like AI driven or not, but on the AI sort of like maybe typical underpinning, so some of the larger, you know, uh, like data structures like, uh, you know, S3 or, or something like that in that, um, there's this combination that I think technology's gotten great at helping provide, which is not only being able to recover at just ridiculously massive scale and doing it really fast, but then Steven, you kind of suggest these, like what we, we call like micro recoveries. It's just these like kind of near real time. They're, you know, they push out, especially in cloud native apps, they're, you know, as, as, as updates are being pushed out and all of a sudden something gets corrupted or, or clobbered to being able to just have these kind of micro recoveries.

I think spanning that gamut is, is I think, a key new requirement for resilience. And so, you know, you, uh, appreciate you kind of giving a nod to us, but, you know, I'm biased kinda working for Commvault. You're right.

Like we have some of our, our capabilities that we have there are just really well suited to cover that, that broad swath of recovery from just the super massive, you know, billions of objects in, you know, record time to these kind of micro recoveries where you're just kind of backtracking to some subset of your, your data state too, to a form point in time. And again, in close to real time. To me, it seems like this, this whole period seems more to me, like when Linux came out, that's, that's when I experienced this much disruption.

So with you saying that, Tim, one of the things, we're talking a lot about the production applications, but when things are in development, it seems like having that resilience to roll back exactly where you need is important too. Especially with ai, when you've got so much time that you're spending on, um, doing some of, especially the inference training or, or whatever you may be doing in your organization, and having several people going at copies of the same data at one time, it seems like there's a great need to, um, to have a place to roll back to so those experiments don't take longer than they need to. I think that's a good pa like kinda a good parallel or like the, you know, the, the way you mentioned Linux, I think the trend that, that I see is that I, no, I don't wanna diminish the importance of kind of on-prem disaster recovery or like, you know, kind of disaster recovery with traditional or operational recovery with traditional packaged applications.

But it's largely a solved problem, like in, in an on-prem world, you know, most, most customers I talk to, like, if it's not, I mean, it's pretty much real time or near, like they're flipping over from one site to another. Like they just got that lick. They, they're doing testing.

It's, it's a really rigorous, well-established process that's not true in the cloud. This operational outages are still an unsolved problem. Um, so, you know, they have cyber and operational recoveries to kind of deal with, and I think you're spot on is that this, this notion of, you know, in development, like, whoa, we just, we just updated a whole bunch of stuff.

You know, it, it might be they're pushing out a, you know, new build of, some of the services might even just be new, you know, or infrastructure updates. And like, we need, we need to rewind that part. Like something didn't go right there.

And, and it's not a, an attack necessarily, it's, it's just a, some, you know, something doesn't match what they expected. And that's, I think that's a still like a largely unsolved problem in the cloud. And I think that comes back to, I keep, I always just have been using, like this meteor analogy is as people get that hygiene in place, you know, then like, here comes AI to just, you know, put the additional pressure on.

And I think that's the, just the massive strain that, that I see within the conversations we have with our customer base. Yeah. I'm interested in hearing more about sort of what the real world looks like out there.

I know that you guys are talking to customers all the time. I've run into those customers, uh, you know, ad events, um, certainly at Commvault shift. Um, what are they saying is the real world of that sort of interplay between AI and data protection?

I think there's a couple things. I think, um, on the fir like it's, it's almost like two sides of a coin is where I spend most of my time talking with our customers. On the one side of it is, how do we make, you know, these new AI stacks, the new data, you know, formats and things like that, how do we make that resilient?

And so this may be things like, you know, it's iceberg or, you know, S3 tables or, you know, Postgres with vector support, you know, so they, they wanna make sure that the, the, um, kind of data stores are something that they can protect and it, and have adhere to their, the policies that they have in place for governance, resilience for all their established workloads. So I think that's one of the places where AI intersects and it's where they look to Commvault to help them protect that data that's ma being made ready for ai. And then the AI generated data.

That's, that's kind of coming out of that. The other side of the coin is how do I use AI to be better at the resilience practices that I put in place? And Steven, you already kind of alluded to some of this, which is, you know, using ai, I mean, it's great at pattern recognition, right?

So like, uh, what about things like anomalies and, you know, detecting those and threat hunting and, you know, helping people find, uh, we use it as part of our platform for our customers to, to help identify these clean points, to recover to, or to help, uh, see malicious activities that are otherwise really hard to detect if you're, you know, a human trying to, to compare this sort of stuff. For sure. So I think, uh, you know, on the one side it's, um, making your data and your, especially your ai, you know, ready data resilient, and then the auto or, you know, the AI generated data making that resilient.

And then the other side is just using AI to improve your, um, efficacy of your resilience practices. How does that strike you, Gina? Again, you're kind of coming from the, uh, customer perspective as well, and from the, the world of, of, of, I guess, uh, tech skepticism generally.

Um, does this, does this ring a bell for you? No, I think it's, I think it's been happening for a long time though, um, within, uh, I think you guys have had some machine learning, right? For a long time and, you know, and so, uh, that's helped that this is not, it's not a new practice, you know, and it's not something new that people have been thinking about.

They've been trying to figure out, okay, this seems like if we use machine learning on this, we could get it to go really fast. Um, and I think it is very true. I think the other thing is you, the people that are doing security and people that are doing, um, the operational piece, which is usually data protection, I think those teams are probably melting into each other a little bit.

And I think there's not a lot of people doing it for the amount of a meteor coming at them and smashing them with AI driven ransomware. So, um, you know, we already know that knocking on the door and getting in is the har is one of the hardest things for the ransomware companies. I'm just gonna call 'em that they're using AI to get better at it, but once they get in, it's really, really hard to detect it unless you've got these tools set up watching for it, even if you're been in the business 30 years.

Because I mean, I remember being assisted men and having little scripts that I would run every morning just to see, and I'd have things set up and we could see when things were happening, but I, you know, it's a horrible feeling to know somebody's dropped a root kit on your, one of your systems and you've been watching for it for, you know, years. So, um, yeah, you have to have, you have to fight fire with fire. And it's not like, you know, it's not like it was 20 years ago.

It's, it's, it's all evolved and it's very, you've got companies trying to infect you and hold your, your data for ransom. So, and they're using AI to get there, so you've got to use fight fire with fire, I think. Yep.

Well said. And I think that that's exactly what we're hearing from some of these leading companies. I mean, when we are at events, uh, with Tech Field Day, um, one of the things that I'm always listening for when companies start talking about, oh, AI is in our product, you know, the thing I'm always listening for is what are you actually doing with it that's productive?

And I love it when I hear companies talk about how they're using, you know, generative AI to monitor, um, you know, massive quantities of data to find anomalies and patterns in that data when they're using AI to, uh, classify and organize and tag data when they're using AI to, as Gina is saying, fight fire with fire. Um, I think that that's great. Um, if, if their answer is, yeah, we've got a little, uh, chat bot down on the corner that can talk to you, well, that's, uh, I don't wanna say it's bad, but it's a little less great use of this technology because, you know, it has a lot of potential and it has a lot of use cases.

And, um, we can have, you know, small specific models that do specific things. And of course, we also have to think about this rise of agentic AI and giving AI applications autonomy and a chain of thought process. Uh, by the way, quick plug, that's what we're talking about on this season of utilizing tech, which, uh, will have just launched.

Uh, that's our, on our other tech field, a podcast, uh, we're talking about agent ai. And, um, I think that that really is going to be opening up a whole other can of worms, uh, good worms and bad worms. So I, I guess to, to sum up Tim, um, tell us a little bit more about the state of the industry and the state of Commvault and what we could expect at Commvault shift here, uh, in November.

So I think, uh, I think it'll come back to some of the points we talked, uh, about earlier, which is I think from a, the state of the industry perspective is the impact that AI is having on, you know, the market. And surely our customer base is not only massive, just, but moving at an unprecedented rate. And so, um, what you should expect it shift is, um, uh, a, a a broad set of, um, you know, presentations, topics, discussions, demonstrations.

We have a, a great both, um, for those who are joining in person in New York, um, November, um, uh, uh, 11th through 13th is we have a set of, you know, breakout sessions. Um, all will have, you know, demonstrations of some of these technologies. And then we have a virtual track, and the topics are going to be around as organizations, um, scale their success with ai, how does Commvault help them bring those workloads, kind of protect those with the sorts of governance, uh, control policy monitoring and enforcement that they have, uh, come to get great at, uh, for the all their established workloads.

And then on the other side, how does Commvault help them use AI to be great at their resilience operations? I think that sounds really exciting. I think it's a really important message and, um, I know with the demos and stuff, you're gonna show people how to do it, so that's always the best.

Yeah, Gina, absolutely. I'm, I'm with you. I can't wait to see what, uh, Commvault has in store at shift.

And, um, definitely we'll be continuing to follow this here on the Tech Field Day podcast, as well as on our utilizing tech podcast. So thank you both for joining us for this episode of the Tech Field Day podcast, focused on the connection and interplay between AI and data protection. Before we go, uh, where can we continue this conversation?

Uh, Gina, Uh, the best place for me is probably LinkedIn. Um, and you can search for Gina Rosenthal and you'll find me. I'd love to talk to you.

Excellent. And, uh, Tim, uh, where can we continue this conversation with you? Same thing.

I'm Tim Zonca, uh, on LinkedIn. You could find me there if you wanted to DM me, but also, uh, for those of us that wanna hear more, uh, get some hands on experience, see some of what we talked about live, either join us in New York City, November 11th through 13th at our shift event, or do it online. And you can, uh, join straight from your desk with good cup of coffee and see the same sorts of things.

So thanks Steven, and good to get, get a chance to talk to you, Gina. Yeah, absolutely. And we'll be, uh, covering shift on the tech field day channels, uh, YouTube, our social media, and of course on LinkedIn, where you'll find me as Steven Foskett.

Also, as I mentioned, uh, we have just launched our next season of utilizing Tech, which is focused on ag Agentic ai and features, uh, friend of the podcast, Frederick Van Herrin and Guy Courier as co-hosts. We've also just launched another podcast that's gonna be relevant to this audience, uh, security Boulevard, which is in concert with our sister website, security Boulevard. Uh, there you'll find, uh, familiar faces, Tom Hollingsworth, uh, Mitch Ashley, along with, uh, folks, uh, like, uh, Fernando Montenegro and Alan Shimmel talking, uh, data security.

So check those out in your favorite podcast feed. Thank you very much for listening to this episode of the Tech Field Day podcast. If you enjoyed this discussion, please do subscribe.

You'll find us on YouTube as well as in your favorite podcast application. I use Overcast. So, uh, don't miss an episode.

Also, please consider giving us a rating in a nice review. We love to hear from you. This podcast is brought to you by Commvault.

This episode is, uh, as well as Tech Field Day, home of IT experts from across the enterprise, which is part of the RUM Group. com/podcast. You'll also find us on Textron TV and in our over the top applications.

Thank you very much, uh, for listening, and we will catch you next week.