Techstrong Gang – May 20, 2025

Hey, everybody. Welcome to Techstrong Gang. We're talking AIOps, which is suddenly, well, hopefully everywhere back in a minute.

Hello, everybody. We're back again talking about, well, all things it, and we're gonna lead off the show with a conversation about AIOps in the wake of some mergers and acquisitions. But let me introduce our guests first.

We have Guy Currier. Who Guy, where are you? Are you still in Texas or are you're somewhere else right now, Actually, I am back home in Austin, Texas.

Well, that's good because, you know, our, our lead story took place in Austin, Texas. We'll get back to that in a minute. Stephen Foskett joins us, once again, appears to be in Ohio.

Is this true? Yes. I, um, stole John Oliver's nameless void from the Pandemic.

So here I am in the nameless void of Ohio. All right, well, you know, it's, it's still the coolest place and every time I turn around, somebody's building a data center in Ohio lately. So there you have it.

Then finally, once again, is the charming Bonnie Schneider joining us once again to talk about the latest and greatest in it and climate control and all kinds of good stuff like that. But I think we have something maybe a little bit different this time out. Bonnie, welcome the show.

Great, thanks for having me. Let's get started. I'm gonna go to, um, uh, guy with this first one, but we saw last week Verana acquired a company called Xenos, and they're both in the AIOps space.

And we have seen also, HPE has been making a strong push into this whole category. And BMC talks a lot about it. And what's interesting to me is they all talk about it, not in the context of generative ai, but that's just a piece of a larger puzzle.

They have predictive AI models and causal AI models and generative AI models, and they say it's gonna take a village of models to get all this stuff done. Because I can't just depend on something that's probabilistic for running my IT ops. I can't be wrong, I can't be right eight outta 10 times 'cause I'll get fired for the other two times.

So, guy, is this world changing in terms of how we think about managing it and age of ai and does it look bigger and broader than just the latest version of Chad GPT? Definitely does. Predictive AI predates the big craze.

And generative generative ai, generative ai, generative AI completes your sentences, right? I mean, like, roughly speaking or, or completes a picture. Um, 'cause it doesn't have to just be words and that can be really helpful with thinking, with thinking through things, but it's not analysis in itself.

Predictive AI is more like analysis. Causal AI is more like analysis. I mean, you know, I, I, uh, I'm really amused, Mike, that, uh, you're so optimistic about AI for once when it comes to IT operations, some of the most sensitive, complicated and, uh, really, uh, organization, potential organization shattering, you know, uh, operations in, in, in a company.

Um, so first of all, yes, I do think it's a new world, um, and I'm not given to making statements like that. Why is it a new world? Listen, um, let's, let's talk, let's start, let's talk about verana and, uh, and, and buying Zen os Uh, you call them AIOps companies, but both of them, zens at least, which I'm more familiar with, you know, far predates.

I mean, they've been using machine language algorithm type thingies for some time, but they really predate the whole AI craze. What's the difference between the two? Uh, ANA is cloud infrastructure and ZENS is more like on-prem data center infrastructure.

They're both observability platforms. Um, so why is that significant? Because they started out separately because, uh, observability and monitoring and, and IT operations in cloud environments, especially public cloud versus on-prem, are wildly different in a number of ways.

Mostly having to do with issues of control. You can't go and rip out a server in the cloud if it's problematic. You have to take other measures and you have to measure other things.

So the proverbial pane of glass and hybrid management and all that other sort of stuff has remained continuously elusive over how long, 30 years, however long it's been. As the infrastructure's changed, become more, uh, not just complex, but more capable. So I asked myself could AI in all its forms, especially in I'll say the a word, a gen, the new a word agent AI could v this be the way out of worrying about single panes of class and thinking more about agents and preferred interfaces, um, helping to, uh, you know, look through these masses and masses and metrics and logs and, and traces and so forth, and helping IT shops do what they've always wanted to do, which is to keep the systems running.

It's just the mass quantity of things that an AI of any kind can look at in any one time and come up with an idea or a recommendation that reduces the firing squad from, you know, eight on eight to two on two or something like that. And shorten decision making times in IT shops make their work more responsive to the business. It does, to my mind, show that promise.

Yes, Steven, you've been around this space for a while. I mean, and when I first started covering this space, there was a lot of cynicism about anything related to ai, and most of the people were like, this thing will never learn our systems. They're all unique and they're different.

Um, but now you talk to people and you get this kind of vibe that says, well, I may not be a whole believer, but I kind of recognize maybe I can't manage this level of complexity without some help from ai. Yeah, that's the thing that's interesting here is, um, I think what we found is, is that there are AI models and, and again, as as guy pointed out, we're looking beyond that sort of, uh, lens of just LLMs, you know, it's not just about chat bots, it's about using machine learning models to figure out trends to spot outliers to identify, uh, problems. This is actually a really good use of, of machine learning.

For what it's worth, uh, we launched our utilizing tech podcast before chat, GPT, and the thing that we were expecting AI to have an impact on was this essentially spot the needle in the haystack spot, the trend, help us sift through all this data and figure out where things are going. And to my mind, that's really the core of AIOps. I, it's, it's using ai, not just LLMs, using AI generally to sift through data and come up with valuable insights.

Now, the problem with coming up with valuable insights, and again, this will kind of segue into our next block when we're talking about Qlik, the problem with being analytics and insights and so on, is that you need the data and having, uh, lots and lots of data to process is not the same function as being able to process that data and spot trends and, and make recommendations. And that's why it makes sense to have somebody like Verana looking at something like Zen os. So a as guy like guy, I'm very familiar, uh, with, uh, Zen os, uh, having, you know, been in the IT space for a long time.

I mean, basically this is a company that was founded by nerds like me who wanted to use open source to monitor their infrastructures. They wanted to figure out what, what devices do I have? How are those devices performing?

There's a lot of deep integration with, you know, kind of old school stuff like WMI and SNMP and stuff like that that, you know, predates the modern IT space. But the, but the idea is collect all the data. Well, when, when you collect all the data and when you analyze all the data, then you have, you know, step four profit.

And that's to me, what's going on here. And, and it makes sense. Every analytics and AIOps provider needs data and they're hungry for it.

And acquisitions like this give them that. I feel maybe I'm wrong, but I think we're approaching the point where the models are almost disposable. Um, you'll have platforms like these, whether it's HPE or verana, or BMC, seem to be able to invoke APIs to call different models as needed for different functions and agents.

And I think swapping those models out is gonna get easier as well going forward. So, I don't know, guy, are we on the verge of disposable AI models? Oh, no, no, you're wrong.

Do you want me to elaborate? Um, I, I think that GPT might give a better answer than chat GUY, but, um, I, I, I just think it's just such early days we don't really know, um, the, the potential complexity of, uh, how to build train models. We're just, we're just starting to recognize the fact that the source training data matters, which is kind of absurd that it's not widely recognized from the beginning.

Um, that, so there, there's one sense in which, in which I, I do see your point, which is that, um, you wanna train on like the kinds of masses of information and data that A BMC might have or, or, you know, maybe a, an open source project as well might have, but then, um, ensure the results are particular and specific to your own environment. So I think it's in the former realm that the more and more and more you push into, not, not so much the generative as the predictive and causal, um, models, um, the more likely that they will be reflective of like, or they'll make better decisions until you reach a certain point where the marginal benefit of additional investment is low, right? I think that's kind of what your point is.

But I think we are also just sort of upleveling our ability to develop and create agents to develop and create apps thanks to these AI capabilities or workloads like these observability platforms. And so there's a lot of human technique that can still be brought into these things. We just don't know where it's gonna go.

Mike, Bonnie, I was having an interesting conversation with somebody who was talking about the cost of ai, and they were suggesting we're gonna need something that feels like finops for ai because we need to track the amount of energy being generated, the amount of, uh, compute capacity needed for all these things. So, um, you know, I kind of look at all this and I laugh and I go, you know, are all these things gonna converge at the end of the day? I think so, you know, they, they're calling it green ops when it's a combination of fin ops and, um, just managing DevOps operations.

So yes, that is definitely a trend, and that's also kind of a selling point for a lot of these carbon accounting, uh, companies that are tracking and measuring energy for AI that, um, it can benefit the, the financial operations in a lot of ways. And then they have to, of course, show that through data. So that is absolutely a movement we're seeing forward.

Steven, will people rip and replace their existing IT management platforms to get to these capabilities? Or are they just gonna wait for their existing vendors to kinda add these capabilities over time? And it's just gonna be in the, you know, an ongoing series of upgrades?

Well, there's a hunger for this. Um, as you point out, uh, companies are gonna be looking, uh, at trying to figure out how to optimize their spend on AI and using AI to do that. And it's kind of a self, uh, referencing thing, right?

Maybe you can use some AI to figure out where you're spending too much on ai. Uh, that being said, I think that the nice thing about, uh, platforms like this that can optimize spend is that in many cases, they can pay for themselves, or at least they can promise to in the sales cycle, uh, because essentially they're gonna identify areas of, uh, well to coin a phrase, waste and abuse that, uh, can be eliminated from IT spend. So this is an area that, uh, we're seeing in the future of intelligence side.

Uh, there's a lot of research going on about companies trying to optimize their environments. Um, certainly there's a huge added, uh, uh, demand for that. Um, I am looking at this number in, uh, in the Textron article here that says that, uh, you know, about 70, 71% of organizations are still, uh, reevaluating where they're running these workloads simply because they need to make sure that they're optimizing spend.

That's exactly what a combination like this will do. That's what AIOps should be doing. And so I, I do, I do think there's a, a, a market for this, and I would like to see it, um, uh, I'd like to see every, every company, uh, investing in this, because frankly, it's good for everybody.

If, if we're not wasting, uh, money and wasting resources, Here's what I worry about, Steven, is, uh, the US reliving yet again, only now in the IT ops, AKA AIOps realm, reliving the easy button approach of, oh, these things are AI enabled, they can make recommendations, let's just throw it in there and we're gonna save money. And that's it. Without this recognition, that must always be in the forefront that human beings should be supervising, running, reviewing.

These are, you know, kind of like all recommendation entrances. They're your dumb buddies, enthusiastic and can read a million lines of whatever really fast. But, you know, don't just let it make decisions, right?

Mm-hmm. I wonder though, and maybe I'm not sure if this is a good or a bad thing, but I think it might happen, might we see a reorganization of the IT team? And maybe it's flatter, because today, um, what happens all too often is when there's an issue, there's some sort of gremlin, nobody knows what it is, you know, and we're all sitting in a room and everybody gets invited to come and prove their innocence in something called the war room.

Um, and that doesn't seem particularly efficient. It actually seems maybe people wind up being pitted against each other. So is there another way to think about managing it altogether, guy?

Or is this just gonna kinda, you know, be more in the sand? Well, historically, what it has done is add functions. So now you're gonna have some function within, that's how it'll start within IT ops, IT management, infrastructure management, this additional function, I almost see it as like, call it ops dev, if you like, instead of DevOps, which is, uh, uh, you know, the, there's always lots of coding going on in these, in these teams, but in this case it's more like, you know, platform and AI management, maybe something like that.

Um, that's how it tends to start. What you're talking about to me is sort of, you know, it's possible, it's possible for, um, AIOps folks to become more generalistic, um, knowledgeable about more domains, more, uh, types of, you know, workloads and support and that sort of thing, because they will have the agents or agents upon agents that are helping them with the specifics in the particulars. But I took it a different way.

I took it more as, um, these are faster ways for existing teams to identify where possible opportunity or problem places are. And so instead of it being a war room, it's a kangaroo court where the two poor, you know, uh, domain members are dragged in and said, well, it's one of the two of you. We figured out that much.

I think I'm the cynic, I'm the cy of this conversation. You usually are. Go ahead, Steven.

Yeah, well, that's funny that you bring up the ag agentic area too. Um, maybe there is a new operations model like Mike is suggesting, where essentially these AI ops companies incorporate, um, uh, remedial agents that can go and change the configuration because, uh, you know, if you wanna be cynical, what would be better than saying, Hey, ai, go optimize our entire a AWS estate. I don't know, I think we're gonna see some dramatic changes in, it may not happen overnight, but I would say that, uh, guy, I think you kind of touched on it, we might be looking at the revenge of the generalist any day now, so hold on, we'll see how this all plays out, but we gotta joke through our next flock.

We'll be back in a minute. Hey, folks, we're back and we're got another one of those spiel reports where some of us go to an event and we come back and give us our impressions. Steven was at a click connect event, and they were talking about agentic AI and analytics and all kinds of fun stuff.

Steven, bring us up to speed. What's the future look like here when it comes to analytics? Well, thanks.

Yeah, we were at, uh, click connect in Orlando. Uh, somebody named Guy was with me, uh, there at that event. Um, and so, so you'll hear from him as well, uh, along with Keith Townsend from, uh, our team and a bunch of other folks from the tech field Day side.

Uh, this is our second, uh, time going to click connect. Uh, it's a great event because it is extremely end user focused. That's my favorite thing about it.

It it is one of those events where the, the team behind it, I mean, there's always a lot of end users and so on at these conferences, and you can meet up with them if you want, but at least the team, uh, that, that I work with at Qlik is always trying to set up opportunities for us to talk to those people and learn from them. They have an AI council, they've got, uh, you know, end users that they just sort of come up and introduce me to. Uh, you know, I met a lot of CIOs at this thing, and it was really interesting to see how these people are seeing this new world of ai, uh, ag agentic and, and, and where we're going next.

I would say that the, for me, the biggest takeaway was as we spoke about in the first segment here on, uh, Textron Gang, it's all about the data companies need. Uh, if, if you're gonna make use of any kind of ai, especially, uh, if you're gonna have AI agents that give you advice or, uh, help make connections between data sets, you need the data. And so a lot of the announcements that we saw were, um, I, I'm gonna say nuts and bolts kind of discussions of integrating data in various ways.

Uh, whether it's in the, as my dad would say, the comes into side or the Gaza side, uh, you know, you gotta have, uh, data coming into your system from various, uh, third party applications. Uh, we, you know, Qlik made an, uh, an acquisition there of a company that that helps to bring data into a data lake and process that data and organize it. Um, we also saw a lot of discussion on the, on the data coming out of the other side where, uh, companies are using various analytics platforms they're using.

Yes, ai, uh, Amazon AWS was there talking about bedrock, uh, as a way to help process data and, and to make these data lakes more useful. That's really the key that we're seeing emerge right now. Essentially, if AI is a data superhero, then you need to feed AI the right data.

It needs to be vetted, it needs to be processed. You know, one of the things that Qlik impressed us with last year was their talk about an AI quality score, uh, or a data quality score that goes way beyond, you know, your traditional definition of data quality. Like is it good data?

It's a trust score. Trust score, sorry. Yeah.

Yeah. And, and so yeah, maybe you can talk a little bit about that guy because that, that, that thing really, um, kind of opens up your eyes to the fact that there's a lot more aspects of data quality and data trust than just, you know, is it Right. And, you know, it's funny, for, for a conference that was really built as being focused on agentic, it was really nuts and bolts.

It wasn't a, a, you know, a fleet of autonomous bots out there doing things to your data. It was much more prosaic. It was much more, uh, I don't wanna say clippy, but you know, it was basically you're building an application.

You know, you, you're not sure how to query the data. You've got AI there to help you query the data. You've got AI there to make, uh, sort of up to the minute recommendations or suggestions about how to deal with the business questions that this data raises.

Um, definitely not the sort of pie in the sky. AI replaces humans kind of messaging that you might expect. It was very much AI as an assistant.

So what, what do you think, guy? Well, I agree. Um, so Qlik click, um, you know, I would encourage people to think of Qlik not just as a, as a, you know, for-profit vendor, which they very much are, they're owned by Toma, Bravo and Investment Company.

I don't think Toma Bravo's, uh, you know, out to, you know, benefit, uh, you know, um, I mean they're capitalistic, right? Um, but Qlik is also a community and has been almost from the beginning, a very tight community. Uh, and Qlik, the vendor serves Qlik, the community really well.

So, so they're customer focused, value focused. They're very methodical. You say nuts and bolts.

That's a great way to put it. So, um, I mean, they got into AI when they bought, um, uh, uh, big Squid, um, that was in 2020. So it, it's before, before generative ai, before the AI craze.

Um, they've made a lot of acquisitions over the last five years. That was a significant one. Their AI strategy is just that sort of nuts and bolts, methodical, not generative AI so much as, um, uh, um, predictive AI because that's their business data analytics.

That's been their business. That's what they focus on. I think though, the big theme was trust.

In fact, two days before the conference, their first press release time for the conference had to do with the, with a statement from the Qlik AI Council, the Qlik ai. I'll get to that statement in a second. This council was announced a year ago on stage, I think you were there too, Steven keynote.

There's four very impressive members from around the world with varying backgrounds. Um, and you know, it had that feel of this big marketing announcement and you know, we're gonna be building trust and all this stuff. But since then, not only have has Qlik itself taken various concrete steps in the trust department, the trust score being the main one, but the click AI council also has been pretty active.

It has not changed membership. And that first press release was a press release from the click AI council saying that essentially, I'm gonna paraphrase, AI is not going anywhere without trust. AI cannot scale without trust, I think is roughly the statement they made.

Trust, meaning that when you use it, you get the outcomes that you expect and you can understand where those outcomes came from, where those recommendations came from, where those decisions, where recommended decisions came from. The trust score is based on the same thing. It has to do with the provenance of the data used, both for training and for, you know, if there's rag or some other form in the in inference as well as transparency about it.

How not, not is it trans is like how transparent is it? So it's not trust in every sense of the word. I think there's a lot of nuance there, but exactly these folks click's core business was data analytics, business intelligence from the beginning.

And they have ex, they expanded forward into AI long before the generative AI craze. And then a couple years back they came backwards with the Talend acquisition that has to do with data quality. They seem to have recognized the issues and analytics and AI that, you know, we keep harping on, but would seem to be fully penetrating in ai, which is the importance of the quality of the data.

And I would like to talk about the olver acquisition too, but I'll just, I'll just stop there 'cause I can see Mike's getting ready to scratch just said next, ask a question. So can I ask you two questions? 'cause there's been a pet peeve of end users for as long as I can remember.

And the first is, you know, you get these analytics reports from it, you don't know where the data comes from. And most people, especially business execs, don't trust the reports they get outta it. They just look at that and they go, you know, something's wrong here.

It doesn't jive with what, how they understand the business. And, and, and you know, we used to generate these stacks of reports that nobody read. And so I'm wondering, is a that gonna get any better?

B is the other frustration was I'd go ask it a question and you know, they'd get me an answer and they'd be like, well, here's your answer. And it would be like 10 days later and it wasn't actionable and I couldn't ask the next question because I'd have to wait another 10 days to get an answer for the previous set of questions. And so the whole thing wound up feeling like an exercise in futility, is this gonna get better?

It'd be great if it got better. Um, I will point out that, uh, your experience is not unusual, uh, from it. And also that your experience rhymes with what happens in the data and analytics space.

So analytics is a sort of another world of, uh, of it. I guess you could think of it as it, but it's really not. Um, unfortunately the truth is that the data analytics folks have been very frustrated over the years because just like what you described, they have a reputation as being sort of inscrutable eggheads essentially.

You know, the business will come to them and say, Hey, uh, I need an answer. Is it A or is it B? And they'll get back reams of data and, and charts and graphs.

And a lot of it depends instead of, Hey, is it A or B? Uh, one of the things that analytics companies are trying to do is sort of democratize access to data so that, you know, they don't have to go through a data scientist to try to figure out the answer, that they can actually go direct to the data themselves and talk to their data. Uh, that's one of those vision sort of things.

Um, and hopefully get actionable, uh, information from that instead of, A lot of it depends. We'll see if that's happening. It, it reminds me so much of AIOps and IT operations, generally IT infrastructure, trying to figure out ways of communicating with the business about what they're doing and what their goals are and, and just, just basically trying to align everything that we're doing over here with what's actually being discussed on the business side.

Um, so, so is it changing? Uh, maybe, uh, that's certainly the goal. I'm not sure if they're achieving that As one of those inscrutable, or at least hopefully former inscrutable eggheads.

Um, I have a different take on what you just said, Mike, and I think I, I'd be curious what Bonnie's take is on this as well. My take is that, um, on the one side, the inscrutable eggheads, um, can't understand how nobody else can understand these reams of things and, and reports and multiple graphs and discussion and stuff that they produce, which is all very accurate and pretty darned impenetrable. But the source of the problem you're describing, Mike, is the fact that business leaders, um, maybe not even even, you know, most of them or, or sorry, all of them, but certainly most of them, um, they actually already know what they want the ANA analysis to say.

And what they don't want is to get analysis that does not confirm those priors. And, uh, there's a real difficulty to stop, look at what you're looking at, ask questions, and listen, especially when you're moving at the speed of business, you know? So that's really been my take on it is if you're not, you've already made the decision and you want the report to support it.

And if you're not getting it, that Venus and Mars miscommunication between the, the data analysts and the business has really nothing to do with any of this technology. Well, Let's get Bonnie's thoughts in here real quick 'cause we're coming up on time. Yeah, no, I think that that's true.

There is kind of a bias where you're looking for the data to support the position that you have, um, with it. And, and it's very easy to do that, that, um, using ai and of course, as the AI gets to know what your queries are, it's gonna do that for you anyway. So, um, I think there is, um, that bias that, uh, we're, we're talking about, but also, you know, having it, um, having the, uh, user and the company and the business to be honest and more transparent in how they're doing it as well.

Yeah, I just gotta say, so often you hear people talking about the bias and AI models and I always look at them and I go, you know, isn't that like the kettle call and the pot Exactly. Goes Well, it goes two ways. It's true.

There you go. All right. Hey folks, we'll be back in a minute with our next segment from Bonnet 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 Techron Group. Well, recently right here in South Florida code remix this uh, conference that was really geared towards developers took off with over a hundred attendees that came in from all over the world.

And the focus, of course is AI automation. And I had the chance to speak to the CEO and the person who really put this whole conference together. Jonathan Schneider of Modern.

Hi everyone. I'm at Code remix in Miami with the CEO of modern Jonathan Schneider. Jonathan, it's so great to have you on.

Thank You. It's a pleasure to be here. I'd love to hear some of the themes and why you decided to do it.

First of all, I wanna see more developer conferences come to the United States, really like high impact, deep technical where people get hands on and do things. A lot of conferences in Europe right now in that vein, but just not so many in the us. And so I live in Miami right now and it was important to me to bring our first iteration of this conference here, kind of close to our home in Miami.

Moderna is really large scale automated refactoring of source code. And so, uh, we heard this morning from Dove Cast and Morgan Stanley that they have over 4 billion lines of source code under management to try to keep that stuff up to date, to keep it modern. To keep it secure requires just a ton of manual effort.

Um, really kind of not fun work to do when you just have to keep kind of going back and cleaning house. So we're automating a lot of that work and, and really freeing developers to do the things that they wanna do, which is build net new stuff. And how is AI coming into it?

Tell me about your processes there. Sure, yeah, I think really we have a deterministic system that's gonna go and predictably make the same change over and over again. But we have thousands of what we call recipes that make different kinds of, of changes and those are hard to discover and understand what's possible.

And AI basically is patient to read through all the recipe descriptions and try to find the right ones and apply them. And once, once it's being applied, it's a deterministic output. So really it's a, the best of both worlds.

So tell me about the conference. You have people here that are from all over, it's global, right? It is, absolutely.

We have folks coming in from Europe, from the us, from Canada, Mexico. These are the ones I've heard so far, but really, like he's mentioned developer focused. We have three tracks, really a developer focused one, a leader track.

And the thing that we're really trying out this year is something called a hack track, where we told attendees before they came, Hey, bring your work laptop, bring it with all your code on it, get security to approve it in advance. Almost like we call the pack for the hack. Get that approval before you come so that when you come here we can do work with you Too often I go to conferences and people bring personal laptops and then, you know, there's not much we can do with them while we're there.

So we'll see what kind of work we can get done. So what do you think is the biggest hottest topic in your space right now? A lot of people are questioning what is the role of a developer in the world way, you know, where AI's also, uh, working with them, how do I use that most effectively?

You know, there's a lot of fear I think right now. And so this is a great time for us to get together as a developer community here and, and talk through some of those things. And finally, you said your company, you're based in Miami or what, what are your goals for modern going forward?

Yeah, we're just continuing to grow. Just raised a, a $30 million series BA couple months ago. We've hired in go-to market functions, continue to hire in engineering, and, uh, just trying to make work closely with our customers and make, uh, them as successful they possibly can.

Jonathan Schneider, CEO of modern, thank you so much for joining me. Thank You. Pleasure to be here.

I also did other interviews with some of the bigger players that were there as well. And I know we're gonna talk about one, um, in Diff Blue because I spoke to one of the executives that came in from there. Um, it's uh, it was an interesting mix of people and one of the things as Jonathan mentioned was, is that people were doing real time, I guess hacking with through, through laptops there.

He wanted to have this to be a hands-on experience, not just listening to lecture tracks, but really getting that hands-on experience for developers and seemed to be successful. Now of course, I, I didn't mention that the location was at a beachfront hotel, so I'm sure people enjoyed that as well. But, um, the people I spoke to were very busy, uh, comparing notes, working together and collaborating and I tried to cover some of that in the video and I'm sure I'll show 'em more coming, going forward.

Alright, So this is one of those, don't look behind the curtain at the wizard kind of moments. So here's the real deal with this thing. Every time you hear one of these AI companies start talking about how they created some sort of AI agent to reverse engineer code, um, what they've done actually is modern has an open source platform that turns legacy code into some sort of representation that can be read by an AI agent and most of these folks who are making these claims and just pointing an AI agent at this particular platform to then reverse engineer and, and not telling you that that, that they're using the open source platform underneath.

Modern also has its own AI agents for doing similar things and is, you know, offering similar set of services and making an argument that says, well since we invented this thing we know it better than anybody else. But I am excited about the idea of being able to do all this because the cost of upgrading and switching things is way too high. And I look at it on two levels.

One is we can figure out how to make existing applications more efficient and save some money. And two is well maybe, you know, I wanna switch vendors. And now it's a lot easier to do that if I can, you know, de configure or reverse engineer the existing code and turn it into something else.

So I kinda like this whole thing and I'm hoping that, you know, the cost of switching is gonna drop to zero, but maybe I'm just being overly optimistic. It's a rare Monday. I'm optimistic a lot of piece to it.

Rare indeed. Not sure what to do with that. Yeah, go ahead Steven.

Yeah, I was actually kind of puzzled by this whole statement about bring your work laptops, bring your work code and let's hack on it here. Um, really, uh, I don't want to be weird, but, uh, ain't no way I'd bring my work code to be hacked on by random people at a conference. Yeah, that seems like a security issue.

Uh, I mean he did say, I mean, to his credit he did say get it approved by security before doing this, which I'm definitely on, on board with. Um, did, was there a lot of that going on? Yes, that's, I I, at least from what I could see because I was, um, shooting the B roll for it and I actually asked some of the guides because they were probably like, why is this, you know, person taking video of us just talking over our laptops?

But that's exactly what was happening. So, um, there was that, that real time, um, collaboration and it was interesting 'cause most of those folks were meeting for the first time that were involved in that. So we'll have to hear how it, how it Went.

So let's the, what's the, what's the idea here is is that, uh, you, you can take, um, some, some legacy code that's running in production, um, and uh, parse it for recoding plus replatforming. Is that what you're thinking, Mike? Yeah, essentially that's what they're thinking.

But I wanna get to this coding thing and sharing my code. 'cause we've hosted a hackathon or two, and I'll tell you, it takes about just the first day of everybody going, I don't wanna show my code to anybody else. I'm too embarrassed, I'm too worried about it.

I don't, I don't want to be criticized. I mean, there's all this stuff that goes on that has squat to do with the tech and it's all about the emotions. Yeah, I think, uh, the tech might have something to say after you're done with that recoding and you post it to a staging environment, uh, Mike, um, it's, it's a no doubt helpful and, and useful to be able to do this.

I mean, just imagine combining this with a little MCP or a to a action and, uh, and, and starting, you know, to, uh, deploy some agents to carry out some of the same things using whatever collection of tools you have somewhere else. I mean, there's a lot that can be done, but there's a, I don't think it addresses the root cause of old code through cause of old code. Yeah, go ahead.

I think this is also a chronic issue because the number of developers that are working on greenfield environments is like maybe 20% if we're lucky. And now the rest of us get, you know, and you show up for work and there's some system that's already been running for the last decade, it's probably got all kinds of turds hidden in it, and you're supposed to go fix and run this thing and, you know, it's like, it's really hard and almost impossible to understand the code. So at least, you know, in a mix of tools like this and a little help from Gen ai, maybe we can actually get in there and understand what that code is, Steven.

Yeah, well that's, that's an optimistic assessment. I think that it's a possible one too. I mean, we've seen this, I remember when we were at the share conference last year, uh, mainframe conference, uh, there were companies, I think BMC was one of them, uh, talking about using machine learning to document and untangle old COBAL code.

And, um, because, you know, it's not completely, it, it, it is actually pretty reasonable to think that an ai, uh, could help to figure out what code is doing, uh, could help to spot bugs in code and could help to document code. I really like that about the modern, uh, concept as well. Um, you know, you've got a lot of Java code out there.

Uh, some of it is older, some of it's newer, some of it's better quality than, and, and, and having AI help you with that, you know, it actually makes a lot of sense. Um, frankly, as weird as this sounds, I might trust AI to evaluate my code more than a random dude at a conference. Um, so there's that.

Yeah, That it's, So you have to say mainframe, I, it was in my head and I did not want to use this such a easy pick in Steven because much as, yes, you want to be able to explainability Titanic huge, especially old mainframe systems, but you're just not gonna wanna mess with that stuff. You wanna explain, you wanna be able to see it, there are opportunities to tweak or what have you. But the whole purpose of it being on a mainframe is to make it, to give it this sort of inviable a aspect to it.

And that, that cuts towards what I was saying before is the root cause of the problem. 'cause it's not just mainframes. You have something that's working and very little understanding of the, the, the effects that it has throughout your systems.

I think unfortunately, and to your point, people are terrified to go take these projects on. I mean, we've all heard the stories about the CIO who lost their gig because they went and hired some global system in integrator to come in and reverse engineer or some application sitting on a mainframe most likely. And then, you know, when the job got started, some bus full of grad students pulled up and they all piled in and moved in for two years until, and nothing ever got done.

And then eventually, you know, the board was like, you're clearly crazy and moved on. And I think there's a certain amount of, I don't know, Steven fear or, or I, I'm just too terrified to touch anything that might be working and I'm just gonna leave it all. Well, Maybe fear, but also, frankly, a lot of people have been burned by these projects.

Like you talked about. There's been a lot of situations where, you know, you call in some kind of, uh, assistance, uh, let's say, and uh, the assistance they're providing is not so great. Again, I would trust an AI to help document my code more than a random contractor of a subcontractor, of a subcontractor from who knows where looking at all my enterprise code.

So, you know, I'm, I think I'm coming around to this. I think I like the idea of AI coding assistance and AI coding de describe. Yeah, and I'm, I'm way more on your and Mike's side than I sound.

I just, you know, it's like the, you take the, take the, you know, take the first down, you know, just take it. It's, it's, it's good. It would be enormously helpful.

Just we're not throwing touchdown passes just yet. All right. Bonnie, final thoughts from this conference of yours that you went to?

I mean, what was the vibe from the people there? Were they optimistic or were they kinda, You know? Yes, they definitely were, uh, they were very happy to collaborate.

One of the things Jonathan talked about was that he didn't feel that there were enough of those kind of conferences in the us, uh, versus other ones, I guess he attended in Europe. So he, um, I think this was the first of the first one of the code remixes hoping it's gonna be, you know, one of many. com article, DBL as a Azure all represented there.

And I saw some other names in, in the DevOps, uh, world that Techron has worked with before. So yeah, I would say it was a, a definitely positive. Of course, I mentioned the location that doesn't hurt, but it was, um, people seem to be very eager to collaborate All, so you, you should just only go to conferences in nice places though, Right?

Yeah, if you're gonna have nice views, maybe it'll be inspiring people Com coming from a conference at a Disney property, uh, on Disney grounds. Yeah, same kind of thing. A lot of people came in.

I think there's a lot of traction to going to Disney with click and, uh, going to Miami Beach with this. Yeah. All right.

Well, as somebody who was out on his way to when conference in Boston, I don't think I'll be sitting on a beach anytime soon, but, but maybe next time. Hey, I want to thank everybody for sharing their insights and knowledge, and I would just share this final thought maybe when it comes to ai, it's like FDR said, all we gotta do is fear, is fear itself, right? Because ultimately, you know, we're cautious and reasonable.

These things are gonna be good for everybody. Hey, we want you all to stay tuned for the next kind of episodes of Textron tv. They're coming up right behind us and I'm pretty sure there's more AI in there somewhere.

See you next time.