Techstrong Gang – June 2, 2025

Hey, everybody. Ai, it's getting real in DevOps. You're watching Texty.

Hello everybody, and welcome to today's edition of the Text Drawing Gang. It's a happy Monday, and we got a lot to talk about, and we got some new faces on the gang. But let me first introduce Tracy Reagan, who's one of our, uh, stalwarts.

Been on the show since almost the very beginning. Tracy is still in New Mexico. Good to see you.

Good to see you too, Mike. Glad to be here. And this is gonna be a fun day for a topic.

All right. Awesome. Also, joining us today is Jack Gold, who's been an industry analyst for as long as I can remember, and I've known Jack for, I'm thinking we're going on at least 30 years, Jack, is that right?

Uh, probably Mike. It's, well, we were both six at the time, so it, it's okay. There you go.

But Jack, introduce yourself and tell folks what you do. Sure. So, as you said, Mike, I'm, uh, an industry analyst.

I've been an industry analyst for about 30 years now. Scary to think it's been that long. gold Associates, uh, looking at the various aspects of enterprise technologies, but also some of the consumer technologies that are gonna be used in enterprise and not too distant future after they become popular at the, uh, consumer space.

And looking at all kinds of technologies, including, of course, ai, but also things like security Cloud, uh, the, uh, focus, uh, we have a focus on ROI and and TCO for enterprises especially. So that's becoming very important for companies that want to make a decision on where they want to go with technology. And, and so there's a, a wide variety of technologies that we look at that, uh, help enterprises become more productive, more efficient, more capable.

All right. And also joining us today for the first time, and for me, I am meeting for the first time is Dr. Stacy Thayer.

Am I saying that right, Thaer? Yes, you are. Yes, you Are.

Awesome. Great to be here. Stacy, tell folks what you do.

Sure. Uh, so I'm cyber psychologist by trade. I am, um, program coordinator and professor at, uh, Norfolk State University, where we have a cyber psychology program.

I have been in the security industry, um, since I was calling 26, you know, attending 2,600 meetings and calling BBSs and hanging out with the, the hacker crowd in Boston, um, back a long, long time ago. Um, but so between my work as a cyber psychologist, and then I also work with, uh, various companies in the security industry, uh, a lot of times doing their events, working with conferences, uh, which is great because it helps me stay on top of, uh, all the latest Trends, get to know speakers and meet people all the time, which is great. All right.

Well, I've been known to pause a theory and ask the panelists whether I'm crazy or not, so it'll be interesting to see if I can actually get certified by you. All right. I, anyway, let's get started though.

Um, there's been some amazing advances in AI and DevOps, or at least they've seem that way. We'll see if they actually pan out. But in the last, uh, week and a half, we've seen enro talk about a new coding agent that can work for, I think they said seven hours.

Uh, at the same time, harness has a MCP server that they've connected up to their CICD platform. Uh, new Relic is talking about integration between its AI agents and the AI agents from, uh, GitHub. So you can create an issue and the GitHub agent or will resolve that issue for you.

And then the New Relic agent will check the work done by the GitHub agent. And then finally, Amazon is out and extended its, uh, AI agents to do, um, essentially reverse engineer applications and re factor them. And it doesn't automatically do it, but it does a lot of the scut work underneath it, and it kind of makes it, you know, more feasible to do that kind of thing.

'cause frankly, as anybody who's ever been involved in one of those projects knows it's a career threatening decision. Tracy, you've been looking at DevOps for a long, long time now. What do you make of all this?

Well, most of the viewers have heard me complain about the fact that AI and DevOps hadn't been, uh, married yet. And why and when is it gonna happen? Well, you know, let's mark the day May 30th.

It is the birth of a new way of doing DevOps. I think it's important to realize, you know, when I, when, when I was selling, uh, Meister products in the, in the early part of the DevOps world, I would tell people about automating builds. And their response would be, well, we don't need that because Jenkins automates our builds.

It was like, no, Jenkins doesn't automate your builds your script automates your builds. And Jenkins calls the build because what was, what is Jenkins and what is our what, our CICD service, they're job schedulers. That's it.

They've always been just job schedulers. They've never really automated anything except the scripts that it calls. You could have used Jenkins to, I don't know, automatically open your garage door and start your coffee pot if you wanted to.

So now we have finally entered the world of AI and job scheduling. So, anthropo, uh, I'm sorry, I forgot the name of the company. Philanthropic is killing it when it comes to AI and, and these new tools, they are the ones who in introduced us to cps.

I got super excited then because there was a potential to get rid of plugins. And now we have the potential to have a seven hour window of job scheduling that we can call and do anything we want with, which means that we have the potential to start evolving our DevOps pipelines to include more work. Because right now, companies struggle, really struggle with going and updating the thousands of, of workflows that we have to update just to add software bill of material, uh, generation, which is holding us back.

So I believe that we are going to be, if we embrace this new technology and start thinking very different about how we can manage a job and schedule a job and what jobs, what, what needs to be included in that job in a new different way, we can now start really moving forward with not just DevOps, but the whole platform engineering space. So yes, today is a really interesting day. And on top of that, when I first heard about, uh, CPS mono contact protocols, I got super excited.

'cause plugins have held us back for quite a while. Uh, the ortel, uh, open source team, they jumped right on it, and they started looking at creating CPS for GitHub and GitLab for doing pull request and issues. So it does not surprise me that harness has built an MCP server around this, because that is exactly how we're gonna get rid of plugins.

So we have, today is a big step in the evolution of DevOps with those two new announcements. And I'm looking forward to building a new DevOps platform around CPS and around, um, philanthropics, uh, you know, workflow, uh, load generation, because that's what it's doing. And I couldn't be happier, to be quite honest.

I really couldn't. Tracy, can I jump in for a second? Um, the, uh, you know, the notion of CPS is great, but because it's, uh, tropic, do you think that there will be some hesitation from other players, some of the major big players to use cps?

As far as I can tell, everybody's keen agreed on a, it's a, it's a quasi de facto standard. What the, the part that's unclear to me, Jack, is there's also talk about this agent to agent protocol, like Google wants to put on top of that. And I can't tell if that's a higher level of abstraction.

I mean, when they launched it, they kind of positioned it as something complimentary to MCP. But I talked to other folks in their kind of scratching their head going, if I have an MCP, why do I need that? So I don't know what your thoughts are there.

Yeah. Uh, and, and that's my concern, right? Because, uh, MCP is great if it get, it's seems to be adopted almost universally, but then you've got somebody like Google coming along, doing their own thing.

Will AWS do something different? Uh, I I just worry that if we get away from standards that the whole notion, and Tracy, you're, you're right on being able to just inter interact with everyone, all the, all the different ai AI systems, AI agents is critical. But if we get away from that, I mean, that's, that's a little bit like saying, you know, in the PC space, right?

Everyone uses USB except you, or USB doesn't fit into my USB, right? And, and, and that, that becomes a real issue. So that's why I kind of asked the question, Well, you know, it's all about adoption, right?

It doesn't mean it's the best of solution. It's what dev will developers grab onto and start working with. And CPS went really fast.

Everybody is talking about it, and, and everybody wants to develop something around it. Everybody understands the impact it has. So developers have really latched onto cps.

So I don't think you know that it's all about developer adoption. It really is. It doesn't mean it's the best solution.

It's the solution that IT developers started working with first and got comfortable with. Yeah. And for all the talk about this great new thing, when you peel it back, as far as I understand that it's A-J-S-O-N based remote procedure call.

It's not rocket science. It's just kind of a, a standard way of doing something. But Tracy, I wanna dive in a little bit more on a point you're trying to make, getting rid of plugins.

Does that also mean well, are we gonna get rid of the scripts too? Because the scripts are the things that are brittle and the things that kind of break all the time and why the pipeline suddenly stops working. So how far are we going here?

Let's pray. Let's, all right, now take a moment and put a good vibe out in the universe for a script list DevOps platform, because the scripts themselves is what holds us back. Plugins and scripts.

Plugins and scripts, constantly backward, compatible issues, constantly updating 'em, can't even update. You know, we're taking, uh, 90 days to update of, of, of CVE because we have to go update all the POM files. We should hope, as I think I said this on the last, um, on the, on last Monday, uh, call, we should hope that a script can be deleted and nobody will freak out.

That should be our goal, because we should be able to have something, something generate our scripts perfectly. Now, I know the question becomes, where is it gonna get the data specific to what you're trying to create? But there should be a way to pass those parameters in if that's what we have to, you know, we have a context window, we can do that.

It's now. So yes, it's, it should be the end of scripts. We should be moving away from these brittle, uh, scripts, um, and into something that's more generated.

'cause that's where the repeatability comes. And that's also where the visibility comes. When you, when you can repeat something, you can more easily fix it.

But scripts across three different applications at the same company can look very different. And when a problem occurs, it's hard to then go and, you know, automate an update for it all because everybody's doing something different. So, yes, we need to get to a point where it's not just the workflow, um, not a, not just a seven hour, seven hour job scheduler workflow, but it should be the scripts themselves.

Stacy, all of this stuff revolves around the idea that we are gonna have these AI agents that are quote unquote teammates on a working alongside humans. Um, are we kinda, you know, as humans psychologically prepared for thinking about AI agents as a teammate? I mean, are we gonna put names to these things and, you know, are they gonna become pets or are they just kind?

Are they just kind of random things? Yeah, yeah. No, great.

Great question. And you know, Tracy, your point you thinking it's all about adoption. It's all about, you know, are are people willing to work with this?

And I was thinking that as you were, you were talking, you know, how excited slash scared slash how much trepidation do we have about a, adopting these AI technologies, uh, and making change? And the idea, uh, if you've ever listened to somebody actually talk to chat beat GBT or talk to Gemini, or you, you thank your, uh, I was in a Waymo the other day. If you've been in one of those automatic driving cars, we all got outta the car and thanked the Waymo, right?

And so I think it, there's a lot about us that, that don't know what we would replace, how we interact with these AI interfaces with, right? And so would they be our best work buddies? You know, maybe would we, we talk to them the way that we talk to our coworkers, what will that look like?

And I think some people are more willing to adopt and experiment and accept, um, than others. So nobody understands me, but my AI agent is kind of where you're going, right? Right.

Which, which you're programmed, right? I mean, if you talk to, to Gemini, it says, okay, here's who you are. And you can program your, your AI buddy to be who you want.

There's, there's programs in psychology or apps of your virtual therapist, right? And is it an actual therapist or is, does it matter if it's real or if the perception is real, right? So, Tracy, how will DevOps workflows evolve in your mind?

'cause I'm thinking about it this way, maybe. So let's say I have 10 agents and you have 10 agents. How are our agents actually gonna get together and actually do something together that's meaningful?

Because ultimately we are all responsible for different parts of the application, or a microservice or whatever it is. But what is that collaboration gonna look like? I don't really think it's gonna change very much in terms of how workflows collaborate, to be honest.

I, I really don't. I think it's just gonna change the mechanics. We're gonna still be, uh, doing the same kind of DevOps work that we do now.

We're gonna collaborate in the same way. We're gonna have workflows that do different things. Maybe a testing workflow, you know, we could have, uh, you know, different, uh, uh, models that we're using.

But at the end of the day, what we're doing is we're making our lives so much easier by taking away all of the broken pieces in the DevOps pipeline. That's what I'm focused on. But I don't think it's gonna change the way we think about the, the software factory floor very much.

It's just gonna change the way we work. It's just gonna get rid of an old process for, uh, you know, job scheduling. 'cause that's really all these CICD tools are, to be honest.

They're job schedulers and some are better than others, and some have historical tracking and some don't. But in the end of the day, it's just job scheduling and we got a lot of logs. So I, I just don't think we're gonna change DevOps very much.

And I've been thinking about this since I've read the article, and I got all excited this morning. Um, I just think that we're gonna do it better. And some of the platform engineering, um, goals will be built into it.

Will it change, Tracy? Sorry for interrupting, but will it change the skillset that people require to be good at DevOps now? No.

I, I think you still really have to have to be a good DevOps engineer. It takes more than just being able to write a good script. It takes a solid understanding of the architecture of an application, um, and how pieces and parts work together and what you need to do to make sure that it's successful when it's deployed.

What we can't see as DevOps engineers or platform engineers is why something broke, because much of the data is, uh, fragmented and there, uh, much of it's in logs in different places. So I would hope in this, in this process of having, um, this AI now ai DevOps pipeline, that we could start solving that problem as well. Um, now what I don't see happening, and we'll see if, if it does in the future, every time you run a workflow, you capture data and versioning, versioning, that information is important.

So you can map that data back to what's actually running in your production environments. So these kinds of steps will still need to be worked with. It just means we're gonna have a better job scheduler, a more efficient one, one that has less scripting around it and fewer plugins.

And that's what we desperately need. I would hope that AI adds the knowledge, as you stated earlier, right? It's, it's, it's all about trying to find the errors, trying to find the inconsistencies, trying to find the differences.

That should be what AI is really good at. Wanna, and that should be give a huge help. I wanna give Stacy the last word on this because as we've talked about on the show many times, cybersecurity is the reason we can't have nice things.

So what happens when an AI agent gets hacked and takes, and somebody takes over an entire process, Right? That, and that's part of the, the question that concerns, especially with AI being such the buzzword now, and I mean, for good reason, it's not to invalidate that, But Do we understand it well enough? Do we understand the predictions of, of how that works?

Right now, everybody is just jumping in to integrate everything, and, and that's great. There's a lot of great good things that AI can do, but we haven't mastered it because we don't really fully quite understand it yet. Um, you know, I'm not too worried about, uh, you know, something going in and going Terminator on us at the moment.

But, uh, but in terms of how we use this, what does it look like? Is this somebody that can take over aspects of your job? To what point does it help?

And then what's the long lasting impact of that, both within an organization, if it goes, if the AI goes down, if it decides to just stop working or doesn't understand the discrepancies in the code. You know, there's a lot of different things that's, that are going on that does require the skillset of understanding how to relate and under and translate effectively. Uh, ai.

All right, well, folks, think about it this way. AI is the undiscovered country, and we're all gonna have this exploration together. And for good measure, that bridge that we just crossed to get there, and we just burned it on the wayside, there's no going back anyway.

So here we go. We'll be back in a minute. Hey guys, we're back.

And there are these new devices in the Zoological Gardens of PCs. They're called AI PCs, and they have these things called neural processor units in them. And, uh, Dell and hp and just about everybody who makes anything is building one of these.

And they're kind of aimed at accelerating or running AI models or inference models on a desktop client. And Dell has one that's saying specifically AI data scientists and developers. 'cause you know, those folks need to work with models locally.

But Jack, we've been talking about the rise of these devices now for it feels like the better part of a year or coming up on it anyway. Um, do we need them in the first place? And BI mean, are they gonna become the devices, everything gonna be an IPC?

Or where are we on this adventure? So the answer to your first question is, yes, we need them and we need them for various reasons. Um, and, and I can get into the second piece because of the first piece.

So, um, AI works best in certain types of hardware. Uh, CPUs aren't great at ai. You can do some simple tasks, AI tasks on CPUs, but you really need heavy duty parallel processing, which is, you know, NVIDIA's game, right?

That's, that's why NVIDIA's doing so well in the AI space, uh, with their GPUs. Now, in the early days of early days, meaning, you know, six to 12 months ago, uh, AI in a PC was really all around, uh, having an Nvidia, A-M-D-G-P-U installed on it, and you could do some inference modeling and inferencing capability as we moved on. Now that, uh, most of the PC chips are including either a built-in NPU, you know, Intel and a MD and, and, and Qualcomm have built in NPUs in their devices or standalones, which are, which are higher performance from Qualcomm or a MD or, or whoever.

Um, we're moving to an area where we can do a lot more AI processing in a short period of time. And so what's happening is we're seeing a, a split in the marketplace, uh, for, in the PC space. Uh, we're seeing a split between consumers of AI and developers of ai.

Let me start with the developer piece. In the past, I'm going back 30, 40 years now. Uh, when, uh, developers first started, um, doing cad, doing a lot of different kinds of graphical systems, they needed to get on high in those days, high performance, meaning many computers that had probably 10 MIPS of processing capability, uh, to do their work.

Then Sun came out with workstations. Workstations were standalone. They were the PCs of their time, basically high powered PCs that let those developers, those designers do their work at a local environment, not having to timeshare on a larger machine.

And there were a lot of advancements made because of that. And then it of course, moved down into the PC space over time as well, where we do a lot of CAD design, uh, on PCs. Today, designers have high powered PC machines.

We're seeing the same process play out with AI today. Most AI modeling is done at the cloud level, at, you know, on a hyperscaler or, or a large in-house on-prem system with lots of Nvidia GPUs in IT. Development is very expensive on those machines.

If you can deploy high powered, high powered for their capabilities, PCs that have the ability to, to, that you can give one to each of your engineers that allow them to do localized AI modeling, AI tuning AI capabilities, uh, that is a much more efficient process. And that's what we're starting to see with some of these new AI PCs coming out. You know, Dell announcing one.

Others have announced that they're gonna move in that direction as well. So you're gonna see a lot more AI flow down into the personal space. You're also gonna see AI running on our machines.

So to answer your earlier question, is this the future? I think within the next one to two years in the enterprise space at least less perhaps. So in the comp compute in consumer space, because that's more price sensitive, most enterprises will have any new machine they buy will have, uh, NPU uh, AI capability built in.

It's just gonna be there. Uh, you're not gonna have to buy it separately. That will raise the price slightly, but not enough to, uh, offset the increased productivity that you're going to get.

Consumers will take a little bit longer. So the AI space is high growth, it's gonna be very important. That doesn't mean there aren't gonna be a lot of lower NPCs that perhaps don't have NPUs or, or gpu, big heavy duty ai GPUs in them, but they'll be relegated to the lower end of the space.

And we're gonna have a lot of agents running out our PCs, we're gonna have a lot of AI capabilities. We're already seeing it with, you know, copilot and things of that nature, but it'll, it'll only get more intense over, over time. Tracy, is this on your wishlist or for all?

I know you probably have one already, but, um, you know, is this on the top of your things? I want to get It, is absolutely on the top of the things that we need to get. So for, for, for example, for us to try to build a, a DevOps, um, small language model, right?

We would go out and we would have to, um, use one of the services like a Google, and it's about 25 cents a GPU per hour, and we're gonna need that running for a month, two months. We could buy some pc, we can buy some local PCs and never have to pay that price again, right? So it, it will help, um, it will help the developers change the way that they develop if this is right, available to them that they've already purchased and they're not having to set up an account in order to do it.

That's how I see it. It's gonna give me the tools that I absolutely need to adopt AI into the software that we're creating at a far lower cost way. Lower cost, because the, uh, you know, and, and I realize that companies like Google probably make quite a bit of money on this stuff, and it's gonna disrupt their, their model, but we need it cheaper and we need it at our fingertips.

Well, Google will still do fine, as will AWS and Azure and everyone else, but they'll, they'll be more, more, much more segmented towards the high end of ai. Uh, the kinds of things, Tracy, I think that you're talking about really work very well on a personal level, on a, on a local workstation. And so there's gonna be a, a division between high-end, very large models, and as you say, smaller models, SLMs, that will work great on some of these new workstations that are coming, uh, online.

And by the way, as, as we all know, you know, the semiconductor model, Moore's Law, they're gonna get more and more powerful every year. So we're gonna be able to do some pretty fantastic stuff on these machines. And A hundred billion parameters doesn't seem small to me, right?

That's pretty, that's, you know, when we first started talking about this, I could imagine that, that, that, you know, that large of a system running on a smaller, uh, on a, on a, on a pc, Well, some of these PCs are gonna have what, you know, 500 flops or going forward or, you know, AI ops is probably more appropriate. So yeah, I mean, it's inflation, right? It's everything goes up Next year.

It'll be puny, but this year it sounds big, right? All right. No, that's actually my question, Tracy.

I, I mean, is now the time to buy an AI ma machine is, or do you think they'll be, do you wait a year? Is it like the next iPhone where you just wait for the next iteration and maybe that one's got a stronger capacity? Or is now the time it's like, yes, this is where you should start.

Should companies be investing in this model versus, say, waiting a year two, three to see how things can grow? You know, I think, you know, like when the internet first came out, I don't wanna be stuck on my 2,400 bo modem dialing up when now there's wireless and, you know, faster, uh, technologies. How long do you think if you had to give a timeline would take before you'd say, now's the time to buy When you need it, when you need it, right?

When you we're at that point. We, we need, we need it. So, you know, we're not gonna wait because we need it now.

That, that's been the question forever with PCs, right? And, and enterprises, do I buy it now or do I wait a year? Because I know there's gonna be a new chip from Intel or whoever, a MD, that that's gonna be more powerful.

But Tracy, you're right, if I need it now, you know, I can't wait a year. I've gotta get the work done now. And, and that's important.

There is another piece of this, by the way that, that we haven't talked about that I wanna address very quickly. And that is that from a personal workstation's perspective, if I'm working in ai, in theory at least, security is much enhanced. I'm not putting a lot of stuff out in the cloud that people can tap into.

It's all localized, or I can at least try to keep it localized. And so, uh, security from an AI perspective is, is a very important aspect. Uh, anything we can do to enhance that.

Uh, honestly, that's the part that concerns me most about AI in general, is how do we know that the models are not doing bad things to us? How they, how do we know they're not hallucinating? There was an example in the press this week about, um, a model trying to bribe people to keep them from turning it off.

I mean, that's getting pretty extreme. We, We, we talked about that last week on the show. Yeah.

Yeah. I wanna Stacy know, going back to why we can't have nice things. I mean, Jack is saying maybe the endpoint is gonna be more secure than the cloud, and that's possibly so, but last time I checked, we weren't very good at securing endpoints in the first place.

So, um, so I'm kinda looking at this going, how many of these things are there? Because if I was a criminal, I would scan for a I PCs and go, those are the ones that are most interestingly target, Right? Right.

And, and, and it comes interesting. There's, there's of course the, the discussions around, uh, humans and the human factor being so susceptible to vulnerabilities and, and being oftentimes the, the most challenging aspect of a, of a security program. And I think what's interesting with AI is that we as humans, we were looking for patterns.

We're looking for predictability, especially if we're trying to break into something or hack into it, how predictable can AI be? If, if we all, if all four of us went in and, and asked the same question or tried to write the same, you know, enter in this code or did the same thing, would, is the output predictable? And if it is predictable, does that then become more susceptible to security vulnerabilities and flaws?

Because it can be replicated or can, where, where does the human factor come in within these AI models? 'cause right now, there's still a very heavy dependency on the human part to be able to work within these functions, to be able to program the ai to be able to set it up. And do we understand it well enough to be able to build a strong security infrastructure around it?

Do we know where the weak spots are? Do we know where, uh, something could be vulnerable? Are we masters of it?

And are we ahead of the game and we are ahead of the, the bad actors who may be able to take advantage of it? And, and of course, there's also al always the issue of, you're right, but there, there's also the issue of can we use AI to hack ai, right? Secu a AI security hackers are starting to use AI to try and find those vulnerabilities.

So it's a, it's a double-edged sword, And we can use AI to find the vulnerabilities ourselves, right? We, you know, we, we should be doing the hacking so we can do the correction. Yeah.

Yeah. And there's a lot of penetration testers of defensive security folks out there who are trying to understand that and trying to work with, okay, what, how does AI impact both defensively and offensively? I do think, though, going back to the, just for a minute on this topic, going back to the CPS that we talked about in the last segment, um, you know, we really never have gotten that good at API security in the first place.

So I think there is a lot of the vulnerabilities around the mcps, and I think that's a, an area that of, of, of security issue that we should be focused on. Just a thought, I'm gonna end this topic here, but I would point out one thing, the future of security might very well be, you know, me watching my AI agents beat up the bad guys, AI agents and vice versa, and then battle it out somewhere in the ether, and hopefully the good guys will win. But we'll see how it all plays out.

Folks, we'll be back in a minute. Discover techron group, the epicenter of tech innovation. We are your go-to for reaching IT, leaders and practitioners worldwide.

Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients, let's revolutionize your tech journey.

Contact us today and tell your story to the world in the most powerful way with Techron Group. Hey folks, we're back and we're gonna have a little chat about digital transformation. We've been on this topic now forever, and a lot of companies, I think there's probably no company at this point that hasn't at some point launched some sort of initiative, but they keep running into the same issues over and over again.

Technical debt or data misaligned goals. Um, Stacy, you have a, a a background in human psychology, and I can't help but wonder at this point how much of this issue is technical, or how much of this is just us and the way we as humans are kind of structured. And the problem is staring us in the mirror every morning.

So I I'm always gonna go on the side of the, of, it's always that person looking back at you, right? Because the technology is great, but as right now, it's only as good as the humans that are behind it, right? And, and the way that we can explain it, and I think there's a lot of, you know, people who are working technology that then they get it, they get in with their teams, and then they go to the boardroom or they go to the executives and they're like, okay, how do I translate this into a good business case, a good business model?

I can explain everything that we need, need on the technical aspect that makes perfect sense to anybody that needs it about why this would be advantageous, how this would help, uh, create more efficient, uh, code or to check and, and you do all the technical things we want it to do. But then how do you then encapsulate that into something that makes sense to people that may not be as technical or that may need to make business decisions, or that may be also be using it and to use it responsibly. Just like we were talking about humans and cybersecurity.

We need to have a good cybersecurity program. Well, now we need to have a, a good cybersecurity program that also integrates AI and best practices with ai, because it does come down to the people and how they're using it and how responsible the use is, and how well they understand it. Mm-hmm.

Tracy, you built your fair share of applications over the years, and I'm sure that they were all intended to be used magnificently, and, but when you present those to end users and it doesn't quite work out the way you had it hoped, or that they kind of just kind of stare at it and use maybe 2% of the features, what goes on there? I mean, is there some sort of disconnect between, I don't know, are developers left brain people and end users are right brain people and it's that simple? Or is there more at work here?

That is such a loaded question, Mike. Yeah, I mean, there is something called, you know, uh, drift in your functionality that somebody think would be really cool, and a developer goes ahead and puts it in and nobody uses it. There's, there's always that problem.

But, um, you know, we used to say, if we can get 70% of the functionality we've done that people really, really need, we're doing pretty good, right? And the problem for companies is keeping up with technology. It is so difficult to keep up with technology.

You write an application and then you have to turn around and rewrite it to meet the new demands. So let's talk about digital transformation shift for just one moment. Most companies are still trying to go through this.

They're trying to get rid of monolithic applications and move to something that's more easily scalable. A lot of what we've discovered, some of it happened during COVID. I was, during COVID, I was driving down the road listening to NPR and I don't remember what state it was.

Um, but they were talking about the fact that they had so many new applicants that their system completely shut down and they couldn't bring on any new applicants until they got more Windows servers. They were gonna have to go out and buy more Windows servers in order to onboard new applicants because they couldn't handle the, the, the, uh, the workload. So if you are a, an insurance company or a bank or anybody trying to stay ahead of the competition and relevant, you're gonna have to take your old applications, maybe add some new functionality to 'em, but most importantly, build them so that they can be scalable, for example.

So you're constantly in this, this, I don't know, this cycle of trying to keep your software not new, new functions, not new features, just relevant for the new technology. And here we go again, we haven't even finished digital transformation and now we're dealing with ai. So it, it is hard to keep a application up to date in terms of the new features that your customers want and keep the platform that it's running on relevant and able to handle the demands of the economy today.

It is a challenge. Sorry for interrupting. There's also another big piece because I talked to a lot of enterprises about digital transformation over the years.

And, and two things that really stand out about why people haven't been as successful as they, they think they could have been. Number one is they start with the technology instead of the business problem. Define the business problem that you're trying to solve first, and then try to find the technology that fits into it as, as opposed to the other way around.

Uh, and and that's very relevant to our discussion today, which is, do I really need ai? Well, what are you gonna do with it? How is it gonna help you?

The other piece of it is, if you can't define what you really need to your developers, how are they gonna develop a product that you want? And so one of the things that I, I advise clients about is, if you really want your developers to understand what you need, take 10%, 15%, 20% of your developers and stick 'em in the real business, uh, that, that you're doing. You know, if it's hr, put 'em in hr.

If it's customer service, put 'em in customer service. You know, they're not gonna be great at it. But if they do that for four or five weeks, they're gonna understand what the problems are and they're gonna develop much better code longer term.

Uh, so it, there are ways to get better at this. I think what ends up happening to, from a lot of, uh, company's perspective and digital transformation is they think it's a buzzword as opposed to a real business process that they have to work at. Stacy, can we do that?

Can we take all our IT folks outta their proverbial ivory towers and stick 'em in the business units and magic will happen? I wish. No, no, I don't wish, but I mean, really, truly, it's some of them, yes.

Right? I mean, I think just like everything, people have strengths, they have weaknesses, they wanna be able to understand a deep dive into the technical components or a deep, some of them are more interested in, in business. I think in general, and why I say I, I wish I'm a big fan for people understanding the role that they play within the bigger organization.

One, it just helps you understand your value and your contributions. So if you understand that you're, you're, you're in a, you're in it or you're, you're a developer, you're doing something and you're in your, maybe you're in a bubble, you're writing your code, but somewhere in that you're also impacting the business. And what you do impacts other departments.

And you know, it's all a connected system. And I think when you can step out and say, okay, well what is, how does the business practice work? What is happening here?

How did the decisions I make impact the business? And vice versa. It allows you to see things one more holistically, and two, then to be able to really understand your, your value and the way that you do contribute and the way that the decisions that the fact that the decisions you make matter.

One of the questions and the concerns that I often have with, with AI is the understanding behind the business decision and the buzzword. And, you know, I know walking the show floor at RSA was like, everybody had to have ai, you know, and I've talked with, um, with investors and asking, you know, what companies are you looking at? It's like, oh, no, they don't say they don't have anything with ai.

We don't even look at them. com bubble, right? Where it was like, you could have a company, but if you weren't online, you might as well have not had a company now that ended up look, you know, look what Amazon did.

Like it was, put it, you, we need, need to lean into being online. You did need to have a website, but at what point are we leaning so much into these technologies that we are forgetting the business? We are forgetting the why, the bottom line and the role that it plays.

Like, yes, we wanna make this investment. This is worth it because it's gonna result in revenue, it's gonna result in product, it's gonna re, uh, result in better security. So we don't have vulnerabilities.

You know, why are we doing it? Not just can we, um, and pull my, my, my Jeff gold here, right? We're so busy thinking about whether or not we can, you know, um, I'll be thinking about, you know, should we, and so that's usually the question that I have is does it make sense?

Why, why are you doing it? It may be the best business decision in the world, and then go for it, but make sure you have the reason why so you can use it effectively. Yeah.

It's gotta be more than this fear of missing out, right? Right. Exactly.

It's ai, AI tech box. We all have FOMO when it comes to technology. I swear I've talked to so many, you know, you talk to some high level exec right now and they're feeling like they're missing out if they're not doing something with ai, but maybe it's just a little early for them to do something with ai.

And so maybe they should be looking at do they really even have a, uh, an issue with digital transformation? Should they be going moving out of their monolithic apps right now? Maybe they should be waiting.

So sometimes stepping back and seeing if it's just FOMO might be a good idea. Jack, when I talk to business people though, then there are whole frustration with it is still pretty high. And they look at it and they go, Hey, you know what, for all the noise in all the years, the productivity numbers are relatively the same.

And the GDP isn't all that different. And they're kind of like looking at it going, where's the math here? Yeah, where's the value?

What, what value does it offer me right now? And, and that's a real issue in many companies. Um, you, you're right, Mike, you know, they've been investing in technology for years.

It's 10% of the budget, 15% of the budget and their sales are going up by 3%, right? It, it, it's, it's not, you know, there's a, there's a disconnect there. I think what really is important for most large enterprises is that you have not just the dev, the, the developers and the, the, the worker bees, if you will, understanding what they need to do, but having the executives talk together, the best processes I've seen is where the, the C-suite actually moves back and forth, where if you have, you know, we're, we're used to having CIOs, CISOs, you know, CMOs, whatever, it's good to rotate those folks every once in a while have this, the, the, the IT director actually not, maybe not run the business, but it'd be a higher level of the business so they actually understand the business.

So there's, and, and, and, and vice versa, you know, maybe the marketing guy or the HR person runs it for a little bit. Um, one of the ways that works, by the way, as well, uh, from a developer's perspective, and we've advised clients to do this, is take your developers and stick 'em on the help desk for a couple of weeks. They're gonna know where the problems are on those applications.

'cause people are calling up and saying, here's the problem I have. How do I fix this? So if you can do that interdisciplinary thing, if you can move people around so they better understand the overall workings of the organization as opposed to their individual silos, it works out much better.

And that's where you get the better investment. That's where you get the, the improvements. Stacy, 20 years ago I asked somebody, why did you get into it?

And he looked at me with a straight face and he said, well, I like machines, cats, and people in that order. Um, are we getting, you know, a new class of IT people who are kind of more engaged with it and people in technology? Or is that just wishful thinking?

I think so. I, I think, and, and part of it also is we're seeing a lot more, uh, the term digital natives, so to speak, right? Like, you know, 20, 20 years ago, years ago, it was, people were drawn to technology for a reason.

It was, I would rather work with technology than people. And then over years now technology and people become more blended. Who knows, with ai, maybe we'll just go right back and it'll we'll v off again, and now it's just technology.

But, um, you know, when I do some of my, my organizational development of burnout is, you know, like a lot of times people don't go into working with technology because they like people, right? They computers can be frustrating. They don't do what we want them to all the time, but there's a sometimes bigger connection than there is for people.

Whereas for others, you give them a piece of technology and they don't know what to do with it. My, my mother, and I'm sure people's parents could tell you that I don't know if this technology does, and some people feel that way about people. Um, that's Why you talk to six and seven year olds, they fix it for you.

They're The best. Yeah. Oh, my kids, they're, they know more than I do every day of the week.

Uh, but yeah, I think now we're seeing where doing it, you're doing, most people one way or another are doing it for their family, at least in, in some way. It may just be, you know, I even do that. It may just be a simple reboot.

That's what I got. Just reboot it. But sometimes that's enough.

Uh, but yeah, I think at this point, as technology becomes more and more ingrained in our life, it also has to become ingrained in our skillset, which allows us to then merge those skill sets and, and say, okay, I have to work with people and I have to work with technology as well, and understanding the intersection of the two. And, you know, there's something too about, yeah, there, the, on the other side, if you look, if you're coming from this, from the, um, the chief level, right? You know, Jackie mentioned that those folks need to learn more and understand better and maybe cross train.

And I agree completely, but you know, yesterday I was, I think it was on Anderson Cooper I think is where I saw it, but the, um, CEO of Tropic, um, Dario, um, ti I think it was, or o Modi, uh, is his last name. He was on, as you know, talking about how AI was going to potentially reduce employment or in increase the employment unemployment rate by 10 to 20%. Because we're gonna have so many jobs.

The jobs that are kind of repetitive and easy are gonna go away. Now, if I'm sitting there listening to it from a c from a CEO's perspective, I'm okay with that, right? Because I want to increase my profits.

So again, fear of missing out, I'm going to say, why are we not doing more ai? What can we, how can we apply AI so we can achieve a 10% decrease in our, in our costs, in our, in our labor costs, and still be, and be more productive? So the, again, as I said before, there are so many, um, there's so many parameters hitting, uh, the, the C-level, a branch of businesses to try to make decisions around what technology to use and why to use them.

It may not be you're gonna get a dev better DevOps pipeline. It may be, what are we gonna deliver at the bottom line? You know, that's what they're looking at.

They're looking at cutting cost. They're not necessarily worried about delivering a better product to the end, the end user. Somebody else is worrying about that.

The product manager's worried about that. They're worried about how, how we can do it at a lower cost. So all of these, um, you know, digital transformation promise lower cost AI is now talking about a lower cost.

This is what's driving many of the execs to make changes. It's, it's, it's bottom line. If I could make about that, Tracy, number one is if from, from a lot of companies that I've talked to that are looking at ai, it's not just about lower costs.

It's about I can't hire enough people and I'm hoping AI will, will rescue me, especially in the development space, right? The second piece is, you know, specifically to the tropic CEOs saying all, all of this stuff, I, I won't use the term I normally use, but it's, it's garbage. Uh, we are very good, everyone's very good at making predictions about, uh, when things will come.

They're usually off by a factor of 10. And, and how impactful it will be also off by a factor of 10. You know, if it's, if he's saying it's in two to three years, it's probably gonna be five to 10 years.

You know, a lot of the predictions we make just, just never, never come true or don't come true for a long time. So we have to take those with a, a real grain of salt. AI may actually eliminate jobs, but we'll also create a whole bunch of new ones.

And, and we gotta, and It's to make, and, and it's gonna improve the lives of people. I mean, DevOps engineers don't need to go have to update thousands of workflow files to put a step in it to generate a, an sbo. That should be, that should be generated.

Let's, so it's gonna make our lives better. But I'm just saying that he's on CNN talking to the wider public right about this topic. And see, and CC level execs are gonna listen to that and say, are we going to miss out on being able to cut our labor?

That's just how the reality Most. All right, guys, I gotta end this. But most of that noise to Jack's point, seems to be for the consumption of Wall Street and investors rather than actual, real, meaningful people.

But I wouldn't leave you with this thought. There's an old joke that says, how do you know when an IT person disagrees with you? They look at their shoes, how do you know when they agree with you?

They look at your shoes. Hopefully that conversation's getting better. Hey guys, thanks for spending some time with us and sharing your thoughts and insights.

And I wanna thank you all for watching the latest episode of the Textron Gang. Please stay tuned for the rest of the Textron TV lineup. There's more programming right behind it, and we'll see you guys again tomorrow.

Take care.