Techstrong Gang – January 13, 2025

Hey, everyone. Happy Monday. Hey, will Legent AI kill the SAS Radio star you're watching?

Textron Gang. Hey everyone. Alan Hummel here from Textron Gang.

Welcome to our Monday show. Happy Monday to you. You know what?

New Year's is over. Get back into the full swing of thing. No one's waiting around for you or going to give you any kinda, you know, easy time off.

We, we are all back to work. Um, and there's a lot going on. We've got a great gang to talk about it today.

Let me introduce you to them. First of all, it, it's kind of a, some of our regular Monday gang members joining us from home in New Mexico, deploy hub CEO Actually, I see you're speaking. Are you speaking at the CVF or doing something with that Tracy?

Uh, I have submitted to speak for the C submitted. I I did get, um, reelected to the open SSF governing board. Yeah, you did.

And we did great. Did, and we did Reinitiate. We initiated a new sig at the CDF that I'm in charge of, called CICD cybersecurity.

So maybe that's what you saw. Oh, That's maybe what I saw. That sounds like G one I may want to get involved in.

Let's talk. No, no, I, I heard that election was stolen. I don't know what's going on.

Yeah, actually I was really shocked. I was like, I can't Believe I got elected on Mike. Count on Mike Debris.

I'm not gonna take the bait. I'm not gonna take it. Tracy, welcome.

Let's move on. Moving into Texas. I assume he's in Texas, though.

I see. Well, there's leaves on the tree. You gotta be somewhere where there's still leaves on trees.

He's future analyst vi co-founder Guy Curry. Hey, guy. How are you, man?

Good. Yeah, I got back from, uh, uh, Vegas in this, uh, CES How C Yes. At one in the morning Last night.

Um, my first CES ever. Mm-hmm. And, um, I will say one thing, it's big.

Yes, it is. Yes. It's, That is a big show.

So I got a cryptic message last night from Lisa Martin at CES that she found my flying car. Yeah, there was a flying car there. Uh, it was a, it was a car, um, that looked a lot like a cross between a cyber truck and a Humvee that had, where, where a, you know, very heavy, uh, it had six wheels on it, by the way.

And a heavy, um, hover, not hover, sorry. Drone like platform pulls out of the back and you attach it to the roof, and then it, you know, it's all these flying cars these days have, uh, have like all these multiple propellers and all that kind of thing. Call, Well, more, more like a, an Osprey air aircraft.

And let's say a Harrier jump jet or an F 35 jump jet look Look more like a drone than a, than a Jetsons flying. It looks like, it looks like a giant drone with a couple people inside. I sure can.

So it doesn't fold up into a briefcase then? No, it, it doesn't fold up into a briefcase. And you don't launch it out of your garage.

Uh, to A point. Interesting. It's probably about a one hour process to get that thing on the back, but it was slick.

I mean, the, the automotive presence there is really big. And so there, there's just so much implication to everything they're doing on that side of the, uh, the sort of consumer side of the fence to everything, you know, we're, we're doing in the enterprise. So it was Good.

Cool. Well, welcome and thanks for being here, guy. Good to be here.

Let's move on there to up to, uh, high up into the mountains, to the guitar man. Right. Mitch Ashley, future of Analyst v vp.

Hey, Mitch, how are you? Hey. Hey.

Good to be here. From the single digit lows of, uh, Colorado. Single digits.

Single digits, man. We've had some That's not like Colorado. Usually it changes.

It. It's, you know, January, that's when we have our cold spell. Hang on.

We have an international audience. We're in the single digits here in Austin. It's just on a different scale.

Oh. So it's just single digit. Okay.

Are you I'm using two single digits. No. Is it that cold there guy?

Oh, yeah. Oh yeah. It was the high today is is in the low forties, but, uh, yesterday the high was around 36 Fahrenheit.

Ooh, It's chilly. Well, It was, it's three degrees in New Mexico this morning. Ooh, you're colder than we Are.

So, well, we were cold yesterday, but it's, it's in the mid seventies and sunny today. It's good to be, it's good to be a Florida Gator. Still in my college football thing from the Orange Bowl.

Uh, moving on from Mitchell though, Mitch, thanks for being here today. I don't know how cold it is though. You know, you can never tell from the flannel shirts you wear every day lately.

It's our Chief Content Officer, Mike Zahar. It's 40 degrees here, baby. And you know what?

It's balmy. It's so balmy. 40.

All right. Well, welcome gang members. Thanks for being here.

We're gonna kick things off today. You know, uh, Satalia Nadella from Microsoft recently did an interview on someone's podcast, and he basically said, SaaS is dead, will be sued, you know, let, maybe a little premature to print the obituary, but, you know, he claims the killer is Agentic ai. Um, Mike, why don't you, Yeah, and we've alluded to this in, uh, some previous conversations we've had about this, where, you know, the debate is, am I gonna use all these AI agents in the SaaS applications provided by Salesforce, or am I gonna go and just have my own and maybe stitch it all together?

Of course, I think Microsoft has a dream where you're just gonna use their agents for everything. And you're gonna turn all these other applications into headless services, essentially, where I'm never gonna see them. I'm just gonna invoke, whether it's a spreadsheet or a CRM app or whatever it is, I'm just gonna call on that, but I'm never actually gonna go enter something into it.

'cause the agent will do that. But Mitch, I know you're tracking this. What's your take?

Well, I think you, it was a really good interview. If you get a chance to listen to the podcast, it's an hour and something long. He's not on all of it.

But, um, essentially the, the SaaS is dead. Is the clickbait coming out of it. He was really saying, I think, here's how I interpret what he was saying is, is the use of AI and use of AI agents is kind of the next evolution of our software architecture.

You can say whether it's a revolution or not. Um, but we've been through this process, right? We do.

We just went, have been through the cloud native transition with microservices and containers and breaking things up and distributing them and orchestrators that was, you know, what we've kind living with now. And the next evolution of that is those microservices, if you wanna think of them that way, or parts of our applications, um, our, our AI agents. They don't have to be software that we've written traditionally or maybe a software written by the agent itself.

To me, it's, it's kind of a whole lot of nothing. 'cause there's a lot said about, well, Excel is just a gooey on top of a database. Well, so is every other application, But that's exactly the point.

He said that all business, virtually all business applications are no more than a database with a gooey On top of it. So my response is, no s**t, it's been that way for decades. When we started, when we started separating the user interface, sorry, it's Monday, so I had to start, it got PG rating.

Yeah, I, sorry, I, I'll bring it back. No, I gotta put a warning on flashing lights in Mitchell. Curing Model view controller.

Have you heard, unless you watch it after ler, if you've been in software for any length of time, we've been separating the user interface from the business logic, from the database access. This is, it's the same thing. It's a new form of that, right?

It's there, there's not any real new news here other than just saying he wants everything to be agents on the backend instead of traditional software. Okay. Alright.

That's, maybe that's news. I don't think it is. So I think it's a whole lot of, it's not a whole lot of nothing.

As in he's not saying anything. He's just saying this is where software is going and where he thinks what it thinks are gonna look like. Is it the death of SaaS?

Oh, heck no. I mean, at, at all of it is gonna be SAS in some way, if you want to think it that way. It's part back end of apps, front end of apps, living in clouds, talking over ai.

We'll have orchestrators for ai. Maybe, maybe it's Kubernetes, maybe it's something else. Um, that'll be managing all these agents.

So I, I think it's just kind of a, it, it's an interesting conversation if you sort of peel away the, the, uh, you know, the clickbait part of it. Um, but it, it isn't spelling the death of software, period. So yeah, I'm an associate, I'm an agreement.

I'm in agreement with Mitch. A hundred percent. Yeah.

I thought the same thing. So let me, let me, I did, Lemme see if I could massage this so you guys can get your head around it, because obviously it struck a nerve with both of you, Right? And Me and guys.

Well, all right. So you can't be so attached it. They're cattle, not pets, guys.

Okay, here, here's what I'm trying, I think here's what Sacha is getting at, and here's how I look at it. Will it stay? Will there be stuff hosted in the cloud?

Yeah, of course there will be. The data's going to be up there in the cloud, the bi. So if you look at apps at their component level, and Mitchell, you said it, you agree with this.

There's the data, there's the gui, there's the business logic. Those are the three components here, right? Those are the three components.

The data's not going nowhere. We're, we're gonna still need the data, right? Because at the end of the day, it's all about the data.

Stupid, right? The goey, I think GUIs are changing. I think, I think one of the things that gen AI and this whole chat bot revolution kind of thing or evolution has taught us is you don't really need a goey.

I think people would much prefer to talk to their computer and tell them what it wants. You know, ALA Scottie and Star Trek, uh, you know, hello Computer. Um, yeah.

You know, so I, I think a goey is the gooeys as we know it, as the experience is I think going to give way to just a much more simpler human computer interface. The business logic is the weak point here, and I think that's what Satya is talking about. The business logic will be controlled by agents.

Now, what I find really fascinating is, for the last three years, I don't know how much wind we've spent or ex exhale talking about, well, what's after Kubernetes? Is there anything on the horizon that can replace Kubernetes? We got something.

I don't think Kubernetes becomes the orchestra orchestrator for AI agents. I think we're gonna see something new and, and these agents are going to represent the business logic in apps, and it'll, this could be a Kubernetes replacement on the horizon. Whatever comes out to do that.

Now, does that mean that SaaS is going away? No, we're still gonna host stuff in the web and access it. Does it mean SaaS apps as we think of them with the pretty, you know, they're not so pretty, but with the interfaces that we type into and all of that is going away.

Yeah, I think so. I'm always a little suspicious about the timing of these conversations. So work, walk this through with me.

Like what's the motivation for saying this now? It seems to me, you know, this isn't a, almost a boldface effort to tank the valuations of some of those SA SaaS companies when basically trying to say that all these things are gonna wind up being features of some AI agentic framework. And maybe just maybe if those valuations tank, we'll start to see a lot of mergers and acquisitions.

So maybe this is an effort to reduce the cost of buying other companies for 2025. Call me suspicious. You're so sign.

No, I, I, I think I, I wanna look at the, uh, Mike, I wanna look at the co I would love to look at the co-pilot roadmap because co-pilot right now is not to my knowledge, particularly a agent and release, but I guess That's where it's going. Think Another thing guy that, you know, the counter Mike's argument. Where does, where does Microsoft make a lot of money today hosting SaaS-based apps in the cloud?

Yeah. Your agents in The cloud, right? So I, so my perspective is sort of a, I think a long, very long perspective.

I mean, Mike, you might remember way back in the day when we were at, uh, uh, zip Davis together. Um, one of the thing, one of the little themes I had at the time was, uh, what we are calling cloud computing today. We're just gonna call computing in 10 years.

And I, I think that played out. I mean, we're not calling it computing exactly, but just everything is cloud, so to speak, unless it's not. And the distinction has to do with use case, if anything.

And there are, you know, so like, it's just, you know, it's just part of, of the infrastructure landscape right now. Similarly, uh, I don't see why SaaS dies, because ag agentic AI takes over. It's just what we call ag agentic ai right now.

We're going to just call SAS tomorrow still or something, if you follow me. It's, it's a way to deliver software as a service. Now the software is hosted and run by ag agentic ai.

It's, it morphs more, it responds to you, it learns all those sorts of things that you expect it to do. But how is it exactly different from SaaS? That's what I don't understand at all.

Yeah, yeah. That's what I couldn't get my head wrapped around either. It's like, why is this not a software as a service?

Just because the technology. It's, that's exactly, it's What, it's, it's Absolute, it's not the software's a service app as we've known it. I think that's the point.

Okay, great. Yeah. And, and I, I do think that the future of copilot is to become an agent, and maybe copilot becomes that agent orchestrator, right?

There'll be multiple agents. I have a different view of copilot. My, my prediction is copilot overtakes the IDE for people.

Essentially a copilot is, yes, is an adjunct additive to an IDE today, the more copilot does or whatever form it becomes, essentially, that becomes your new IDE, especially for non-developers, if you will. But that's the orchestrator of your development process. Creating software, creating bot, managing bot, whatever it is.

I think that's kind of where it's heading, at least in my view of it. Maybe I'm wrong. We'll see what happens.

We'll see, I Dunno, that's, everything gets going. Microsoft describes Azure as quote the world's computer. This is the same Microsoft that it's always been.

Every time it looks at something, it says, here's this one unified thing that we're gonna own end to end. And everything else is, you know, get out of the way or be absorbed. We are in a hype cycle, and we have to consider a lot of what we see coming out of these, the, what I like to call the giants hype.

Uh, it just, it, it happens for every single time we change technology. Look at all the hype that was around Kubernetes, and rightfully so. We had a lot to learn, and we have a lot to learn about AI and, and how we're going to apply it.

And there may be some really interesting outcomes of that. In in particular, you know, distributed databases where there's data across smaller chunks of data across organizations or even open data that we can, that these AI agents can get information from. But it's just an evolution of software.

So I guess that's where I was struggling. And yes, I would love to ditch the mouse. I'm so tired of my mouse, but I still want my keyboard.

I hear you. Here's, here's, I wanna, I wanna, I wanna say, I give my interpretation of what Satya was saying. What he was saying was that right now you buy a set of, of, you know, soft a set of software, some of which are SaaS that carry out certain functions for you, for, for your job function.

What he's saying is that's gonna be replaced by an, you know, an AI agent or a package of AI agents that are for you in your function, and you as a person and how you carry out your work. And those will go and get the necessarily necessary CRUD functions and put them together and change them as you, and learn and grow with you, so to speak. So instead of buying SaaS, you're buying AI agents.

I think that's what he meant. But that is, I agree with the panel here, a distinction without a difference. I think we need to introduce a new term, which is agen hype Building up one piece.

So, on, on that note, hype ai. Let, let me, let me tie a bow on this one and we'll break, um, today's Monday, right? I do Monday, January 13th.

I do believe that 2025 is going to be the year of AG agentic ai. I think we've gotta stop looking at AI as a monolith and, and recognize that there are, there are divisions and, and features within the AI monolith and agentic AI is one, and it's gonna be huge. But if you're really interested in what 2025 has in store, tomorrow is our Predict 2025 session.

I'm doing a great panel with a bunch of C-level folks on what DevOps and AI have in store in 2025. Mitch, I know you, I saw you, uh, yeah. Doing your session the other day.

I'm my top 10 predictions, top top 10 predictions. And that's good. That's both.

We'll replay that during Christmas next year. And the s we'll do backwards. We do David or something.

Guy. I think you have a session too. No, I do You want to say anything about it?

Yeah, it's, uh, it's work With me guy. My, my predictions, the 2025 will be the year, the fragile app, The fragile app ex, My prediction is my AI agent will be able to beat up your AI agent. That's what I, That's part of what's gonna make my app fragile.

Okay. com right now. Register, it's free.

I've got Daniel Newman talking about stuff. We've got Steven Foskett, so a bunch of the FU folks, but we've got people from, I think, Google, Microsoft, bunch of the great DevOps people. Good cyber stuff.

Don't miss it. Predict 2025. You're watching Textron Gang.

We'll be back in a minute. Hey everyone, it's Sunny And Cher. Ladies and gentlemen, tech enthusiasts and future Gazers gather round the biggest, boldest, most mind blowing predictions for 2025 are coming your way at the Predict 2025 virtual event on January 9th.

Oh, sunny, you're predicting something again, last time you tried this, you said laser dip for the future. How'd that work out for you? Hey, hey, Cher.

Not every prediction a hit, you know, but that's why we've got the real experts this time, top analysts, visionaries and tech leaders sharing what's going to rock our world in 2025. So, no sunny predictions this time, no flying toasters making a comeback. Very funny share.

But seriously, we're talking AI breakthroughs, cybersecurity game changers, the future of DevOps, cloud Innovations, and so much more. And what about my favorite prediction? A smart mirror that tells you how fabulous you look every morning.

That's real innovation. You're already ahead of the tech share. But if you want to hear the really big stories in tech for 2025, you've gotta tune in on January 9th.

The event kicks off at 8:45 AM Eastern and runs until 2:30 PM And the best part, it's all virtual. No stuffy conference rooms, no long commutes. Just grab your coffee, your laptop, and join us from anywhere in the world.

Know what else? It's not just predictions, it's insights, strategies, and a whole lot of fun. 0, predicting your jokes before you tell them.

That's a good one, Cher. But the real joke is if you're missing out on this, so don't miss predict 2025. That's right.

Mark your calendars. January 9th, 8:45 AM Eastern. Be there.

Or you'll miss the biggest scoop on the future of tech. See you. I predict 6 20, 25.

Be there. All right, folks, we're back. And we're talking about AI and DevOps again, because there's two reports last week.

One from IB M1 from harness, both highlighting that, well, maybe all the things are not quite cracked up as they were hyped. We're back to that word again. Um, the issue seems to be that the AI tools are generating code without any context for the production environment.

So the code doesn't run, and then people spend more time debugging the code than it would've they had actually written it themselves in the first place. At least that's the theory. But Tracy, I know you've been kind of following this all last year and we're coming into this year and it, what is the fundamental issue that we need to kind of work through here to make all this AI stuff really work?

Well, I think first we have to separate out DevOps, to be honest. Um, I would say Dev has embraced ai, you know, with tools like copilot. Um, I've, even myself, I use, uh, chat GBT to generate my markdown from my websites.

There's quite a bit that, that we do on the development side. But to be quite honest, DevOps hasn't really changed that much. Um, and there may be tooling out there like, you know, AI and, and testing that can be updated in the DevOps process.

But I think this, this, you know, this survey and this article and this information is spot on. We, um, we have, I feel I've thrown DevOps, uh, DevOps out the door a bit. And to say that we're doing DevOps and ai, is it throughout the life cycle is, is simply wrong.

We're not, so when we see these surveys, I think we have to, you know, I participate in surveys when they come across, but it, you know, it states that 99% are using coding tools infused with ai, which is probably right. But few of them are implementing any kind of really solid a, uh, AI practices in the DevOps pipeline. And I, you know, I, I'm very black and white about this.

There's no gray for me when it comes to a Jenkins workflow. And have we changed anything in that over the course of time? Not really.

We have not evolved the core process of our software factory floor. Uh, so while we talk about it, there is a lot of hype in what we talk about. You know, just being able to use AI to go out and find a, uh, somebody to approve your code isn't enough.

It's really not enough. And most of the DevOps, uh, engineers that I speak to, they don't have a lot of experience, um, in you, in what we need to do in the DevOps pipeline to start catching these kinds of problems. Um, we rely on the, you know, somebody to check in code and then approve it and push it across the, the DevOps engine takes, uh, starts working.

Is it gonna find code that wasn't really written? Well, I don't think so. It could scan it and find out that it's got, you know, it's pretty clean, but does it work Well?

I don't know. So I feel like we are in this hype cycle for ai. A lot of these companies want to get out in the front saying that they're doing AI and lifecycle management, but it's not happening in reality.

I don't see it, and I'd love for the panel to prove me wrong. I think that's exactly right. I think that, um, you know, we, we, we, we, AI is an, is an assist in so many helpful ways, um, to get you started, to get, i, I call it a reliability and a quality boost.

But the quality part requires your intervention review. Um, if what you're aiming for is speed, then I think, uh, you're just gonna get more garbage faster. And that's what we're seeing.

Um, if you disconnect that speed, you know, productivity, if you like element of it, that really honestly gets all the focus. But if you take that out and you realize that you can produce better annotated, uh, you know, better, you know, better functioning, um, and, uh, uh, more reliably produced code that still runs through the same processes that you already had, um, then you're using it well. And I think that, that, you know, the developer community as a whole has not this, and I, what was not as clear to me, um, although you have pointed out to me a few times, Tracy, is the stress it's putting on ops teams for whom the, you know, the code is, the code is the code, and, uh, nonetheless, they have to deal with the consequences.

So it was very illuminating in that sense. One thing I noticed also though, was, um, that this survey covered, um, both the coding side and the incorporation of services side. And that's a whole other reliability, um, uh, question, right?

Um, which is incorporation and use of, of services, which can and should change over time, as pointed out in the article, um, that creates its own set of instabilities and its own, if I may say so, fragilities, um, that, uh, uh, can affect the, the app, um, the app app performance and outcomes. I see what you did there, Greg. I think we're still at the, I'm sorry, Mitch, go ahead.

We're, we're still at the individual productivity level with AI and development. It's a copilot, it's an augmentation to the work. It's maybe there's some mag magenta functions that it performs, you know, software quality teams still have to treat all code as suspect, whether it's AI generated or not.

They kind of have to not care. Everything needs to go through the, the quality process. And, and to your point, Tracy, it hasn't made, its, has not, has not made its way into the workflows, into the pipelines, into the CICD into the processes underlying it.

At least not yet. Maybe they'll have start to happen in 2025, but I think, and maybe it's a agent AI or AI agents that will help us start to move beyond just helping Mitch as an individual developer or helper help, uh, you as a, as an individual software tester, whatever it might be. But we haven't changed the underlying framework of how DevOp works.

And, and it's a lot to disrupt. I mean, don't, don't get me wrong. It's not like, oh, let's go change this.

You know, to your point, Tracy, there's a lot of Jenkins and CircleCI and everything else CSCD pipelines that you don't wanna touch, because when those things break, it's not one of 'em, it's dozens. And how, how do you, how do you de diagnose that and fix it? So it's definitely a place for some disruption, but I think you have to look at it as productivity of the process, not productivity of the individual.

So, Mitch, to that point, I'd love to get your thoughts. You know, I was talking to Nick Durkin, who's the field CTO for harness, and he was of the opinion, and admittedly this is biased on his part, but he was saying that the AI agents used for coding need to be aware of the AI agents that will be used for the ops side of the equation. And that you need to have a cohesive framework for those things.

Otherwise, you know, you're gonna have scenarios where the AI agent used for the coding tool is gonna be working across purposes from the AI agent used for the CICD and the op side of the equation. So I don't know if we need to rethink this or, you know, can all these AI agents find some way to be compatible? Well, maybe that's kind of where guy's going with his fragile app app idea.

But you know, it, it's everybody's, or every AI agent doing whatever it's doing in its own context, limited world of course, could very easily work across purposes, right? My agent, AI agent can beat up your AI agent. One of the things that helps is not the silver bullet, but is that context knowledge graph, right?

So that I know well, by changing this bit of code that affects these people in operations or security or whatever functionality, it's the, it's the broader context of what's happening. And I don't know that we've got sophisticated enough knowledge graphs to solve all the problems that might come up from, you know, AI generated code. But that's at least the start of what we're using to try to add more context than just a single user.

I Tell, I think It's a terrific point. This is true. Where does context come from in these, you know, uh, is it come, does it come from the people who train the AI in, in a certain sense?

Like you are talking about two different context lobes, dev and ops. Where, where at the point of prompt does the context come from other than the user? Tracy?

Well, I, you know, in this discussion, I, I just wanna create a re like a use case. My niece, um, told me when I asked her about she's got a new job just outta college. She told me once she could generate 6,000 lines of code a day, but she couldn't figure out what was in it.

And in this, uh, in this harness survey, it says in here, I'm just reading that 67% of developers space spend more time debugging AI generated code, which is kind of what she was telling me. But even more concerning 68% spend more time resolving AI related security vulnerabilities. So I think it's sometimes it's, you know, we more code is not always better.

And, and in these AI generated environments, these AI where the developers are starting to use AI to generate their code, they have to get better at learning how to use it. What she told me was, she's starting to stop. She's not using it as often now.

I just talked to her this over Christmas. And what she's learned to do is be more specific in using AI when she gets stuck on a particular function. So she's trying to minimize how she's using AI because of the issues that she found with having to debug it.

Because she said, if AI generates it, I can't debug it as quickly. I don't understand it. So it's harder to see what's what it's doing.

So that that's, that's a very real use case for somebody who's just outta college, got her first coding job and having to learn to use these tools. So it's kind of like prop training, right? How do you get better, get better code generated?

Because you're not asking it to do everything. You're only asking it to do smaller functions. Now, if we, if teams get better at using ai, then maybe these numbers will come down.

But if right now we're relying so much on AI and tools like copilot just to go out there and generate this stuff, we'd better be running some good SBOs and understanding what it's pulling in, understanding the licenses, if it's, you know, if these tools are, are not approved, um, across the organization, we have to understand if they're, if they're compliant within the organization. So there is so much more to do that AI could be solving for us than just generating code. And it kind of goes back to some conversations we had last year about the ghost developer, right?

Uh, and I, and I complained, I said, it's not just about coding. There's so much more that's done in software development that we're not trying to use AI to solve. That could be done.

And I, I think that we, we should be seeing maybe a prediction for 2025 is that, you know, AI ops we're, we'll start having more conversations about AI ops and what that is. It is using AI To that term being corrupted, that term's being corrupted for Ml. I know, I know.

But it needs to be clarified because we need to learn how to pull this stuff into the DevOps pipeline, um, and start fixing it. So I've got some thoughts. I've, I've listened to you all here and I got, you know, I, I should have been taking notes 'cause my brain won't remember everything, but I got a lot to unpack here.

So first of all, I mentioned before about the need for an AI agent orchestrator. That's exactly Mitch, what you're talking about, guide that you're talking about, right? My, my dev AI agent doesn't really jive with my ops AI agent and it doesn't really jive with my security AI agent.

And so they're working at odds and loggerheads to each other. That's a temporary phase. And that too shall pass, right?

We're going to have to have some sort of agent orchestrator that makes sure these things are in concert. Is that gonna pass like a kidney stone? Just wanna make sure I Understand.

No, no. It's like your grandmother told you when you were younger. Don't worry, that too shall pass.

It's just a phase. Um, you were experimenting, but anyway. Yeah.

But so that's number one. Number two to say, you know, what challenges and what failures or, or successes we've had in AI and DevOps is like looking with a very tiny telescope at Uranus or Neptune and trying to tell me if there's oxygen there or gold there or something. Trying to figure out what's going on in that planet from a small telescope here on earth or even a distance star.

You, you don't see the, you don't see the features, you're seeing a blob, right? But yet when we dive in beneath the clouds to the individual pieces, we are seeing AI at every step of the way. We are clearly devs are using ai, are they overusing it?

The, the, the experience of your niece, Tracy. Again, that's a phase that'll pass as AI gets better, right? You're gonna have less security bugs.

It, it's already measurably better than it was a year, year and a half ago. Right? Which is why it's being used more so, and it'll continue to get better.

I, I think the other thing is you haven't mentioned platform engineering, and I think that's a big story for 2025. It was a big story in 2024. com, check it out.

It's one of our sites. I think that's where the agents for ops live. In many ways.

Platform engineering is more about the ops than it is necessarily DevOps or, you know, they, there, there's not platform engineering didn't invent a lot of new things. It reused a lot of old things. And so it's sets up that environment for the devs to go faster.

And so is platform engineering using agents or ai, I think we're gonna see more and more of that in 2025, right? And I think the role of platform engineering in ops is going to be a huge story. Right now we've thought of platform engineering and how it's helped devs, right?

'cause they don't have to build their own environment, but it's really about ops best practices. So I I I think that's another thing in here. I think it's gonna be a cultural issue as much as a technical issue.

Mm-hmm. And my prediction is that a year from now, we'll still be talking about this same brawl. Absolutely.

But it'll be better. One other thing I wanted to mention, Ja Chase, you mentioned Jenkins. Look tomorrow at Predict 25, another subtle plug for Predict 25.

We're gonna announce the winners of the DevOps awards. Now those good folks at KPMG have given me the Silver briefcase. So I know who won already.

KPMG didn't do it, but just, you know, I I I recently watched some awards, oh, the Golden Globes and they had the KPMG guys. But I've looked at the winners for the DevOps dozen awards, I'm announcing them tomorrow. You'd be surprised what's old is new again.

And some things don't change in terms of what DevOps tools and CICD tools still rule the roost. And they're not what you would think of when you say, oh, what's AI empowered? But yet when you look under the covers, they're using more AI for testing both security testing, quality testing, security and quality can be the same thing.

Load testing, unit testing. There's AI being used in deciding what my test coverage is, what tests run, how tests run in, in moving things along the pipeline. So again, when you look at the, the planet through the telescope, yeah, you may not see it, but when you, if you could dive down under the clouds and get to the surface a little bit, there's all these little things going on that are using ai and at some point it, it, it just dawns on you and say, wow, it's all over the place.

It's, it's ubiquitous. And I'm gonna leave us with that word ubiquitous. Um, let's take a break on Textron gang.

We're gonna come back and we're gonna talk about who's gaming the GitHub stars you're watching, texture and Gang Modernize your business to fuel innovation and elevate customer experiences with the builder community. Hub AWS and its partner network provide essential tools for transforming applications and infrastructure to fully leverage the cloud. Discover free trials, in-depth demos and essential resources to empower DevOps engineers and developers to deliver value faster and more reliably.

Visit the builder community hub to learn more. All right folks, we're back on the C block and there was a report that came out over the holidays talking about how the bad guys have figured out how to game the GitHub system. They are creating, um, code sticking in a repository and figuring out how to give themselves a whole lot of stars that entice people to download that stuff.

And surprise, surprise, it's loaded with malware and we've seen GitHub kind of evolve for all the good things it does. It's also become a distribution network for malware. So I feel like something needs to be done here, but Alan, what's your take on what's going on here and how do we kind of make this work in a way as intended?

Well, you know, this is not a new problem. As you said, Mike, this is, this is the developer dilemma akin to people who believe Amazon reviews, right? You, you go to Amazon and it's got, oh, this product, is this product better than that protocol?

This product has 456 reviews. You know, I'd love to be able to see what IP address put up 435 of those reviews, uh, because it's probably the same IP address and it's just people gaining the system. But, you know, why is this not a new problem?

Look, I remember having this discussion five years ago about docker repositories, right? If, if, if the docker image, you know, container image to do, um, I don't know, to, to do, you know, a very common call that would be in one container, uh, a mail call or something, right? Or a send mail call.

And then let's call that docker image was called dot send mail. Someone would, would put up a docker image that was dot 10 mail, right? And then someone sloppy would not see the tend for the send and hit 10 and boom, they got, they got eaten up, right?

It was full of malware. It's a very similar kind of thing combining sort of the Amazon, you know, review gaming with another way of, of GitHub's a repo. It's what it is.

Like all the other repos and there's a lot of bad crap in our repos. The answer to me for this issue is we need repo firewalls. Sonatype has talked about implementing one.

Mitch, as you know, I know you covered Sonatype a lot. They have one. We need a firewall in front of every repo that before I download anything from a repo, it is checked from malware.

It's either in a sandbox and opened up or something happens and it's checked from malware. It's not a GitHub issue though. GitHub is unique in their star system.

Maybe another way of gaming the system. But we need repo firewalls. I can't, can't even begin to fat fathom how much safer and better off we would all be if we had re if we had firewalls in front of every repo that was checking every package that gets downloaded.

I'll leave it there. Yeah, when I saw this article, I just, I I, it frustrated me so much because as a, you know, we have an open source part of our product and we worked really hard to get the 300 stars that we have out there. Probably, you know, talking to people, getting their opinion, asking 'em to go out and look at our code.

So it's, it's, it hurts, right? Just no more than it hurts. An Amazon seller who works, works really hard getting their product highly, highly rated.

I know. Or a book an author or something. I know, you Know, exactly.

But there is a part of the human brain. I'm reading this very interesting book called SAPs. Um, I highly recommend it.

And it talks about the, our brain having been kind of evolved around two things. Gossip and fiction and how we choose to believe whatever we want to believe. So if you see an open source project out there that's got a hundred thousand stars, you know, and you look at Kubernetes and it's got probably close to that, Jenkins probably has half that.

You might wanna question it, but maybe we don't because we believe what we want to believe. And so this is why this continues, because fake stars is like fake news. If we wanna believe it, we're gonna believe it.

'cause we love gossip and we love fiction. It's a human nature. That's interesting.

I like that. Tracy. I think there's a, I think there's a safety factor too.

Like I think about my behavior on Amazon of, well this has got 3000 stars and it's ratings and it's four and a half and this has got 17 in it's four and a half. The 3000 must be better, right? More is more is better when in fact it's probably worse.

Who knows what it is. I think there's just that natural tendency that, that whoever is faking the, you know, gaming the system to get more malware out there, it's just playing upon people's natural predilection to say, I need a quick, easy decision if unless guy recommend it to me and I have to go out and figure it out on my, on my own, while I'll just look at those ratings and go, what's sort of, what looks like the best? I'm not gonna research it, right?

In most cases. Well, But, well, we got two perspectives here because we got Tracy who's like, people will believe what they wanna believe. Well, people are pre-wired to believe what they wanted to believe Pre rewired and to do that.

And then you got Mitch's, which which boils down to can't believe anything. So I'm just not even going to do that. I'm gonna, I don't know if I'd recommend calling me, but I'm gonna, you know, check with someone and, and, and, you know, and, and find out.

And I, I just, you know, I I scratch my head Mike Vard, like when I see this sort of thing and just ask myself are, are we just in an age now where the, the, the BS is just accelerated because I think this BS existed a hundred years ago, 200 years ago with pamphlets of unknown prominence. I I, you know, I think I've used this saying on this, on this, uh, show before that the truth always wins is not when it wins, but not whether it wins. It's when, and, uh, folks using these high star reviews, it may not even be a breach.

It just could be somebody who's gaming the, the system so that they get more popularity. And if it's crap, it's crap. And if you are, uh, uh, using these repos or, or any open source source that's highly rated even by guy personally, and uh, and, and it doesn't work out for you, at some point you're gonna pay, What is it?

Half a statistics are fake and the other half are made up. Yes, You're welcome. Exactly.

Live. Damn. That's what you come to me for God.

But can you show, show me how you do that? Scratch your head, Mike Ard like, okay, there it is. She's, I think what we have to go back to, you know, um, John, I always bring this up 'cause I think about it all the time.

John Schwartz used this term authoritative bias. Mm-hmm. And stars are authoritative bias, right?

They, so we have, we have to learn to question authority, But, but that they've forgotten how to do that. That's a double-edged. So two things.

Number one, how has Amazon try to un game that system? They've now have reviews that are by a verified purchaser. You've gotta be someone who actually bought this.

And there's ways of gaming that obviously, but they're, They're reviewing their reviews, Alan. And so then the authority you're trusting is Amazon, well At least, you know, was a verified purchaser. But secondly, look, this is a society wide scourge that affects much more than our software and malware.

This is why many Gen Z and I, I speak to a lot of Gen Z young people, my kids' age, they don't believe a g*****n thing they read on the internet or see on tv, especially if it's from mainstream media. They think mainstream media is more likely to tell them wrong or lies or non-truths than some dude sitting in his mom's basement in Texas and, and podcasting once a week. Him will believe.

'cause what does he have to lie about? Who knows what he does when he is not in his mom's basement? But thi this is, this is a major problem in our society is, you know, the, the ability to separate veracity from, from nonsense, from lies, from, and in this case malware from people that's gonna, it's the ability to separate Russian propaganda from, from truth or Chinese problem or whoever your, the scourge of the day is.

How do we do be? And and I think it goes back to what Tracy said is to a certain extent, we're wired, we're wired for clickbait, we're wired for gossip. Yep.

I don't know what that answer is, but, You know, we're talking 40,000 years ago, we have a lion headed man who may have started a religion, right? Mm-hmm. And we believed it as omo sapiens.

It's a very fascinating read and it, it, it, it answers so big questions for me right now. And I wanna point out that, you know, the CNCF used to use stars for, uh, qualifying to get into their incubation programs for open source. I believe they've stopped doing that though.

Maybe they recognize something that this article pointed out that I wasn't aware of before. Absolutely. Maybe it's like the old loony tune.

The stars are really the, just those things circling around your head after, well whoa, whoa, whoa. Well that, that, that's all folks. What's a good way to end it?

Um, hey, what a great Monday conversation. Chay guy, Mitch. Mike, thanks for joining us here.

Thank you for joining us on Text Drunk Gang. As usual, we've got Text Drunk TV following, we've got a lot of great text, drunk TV stuff. If you're into knowledge graphs, I think one of the interviews I've done is gonna be playing Right During Today.

It's about a company that's working with not with, uh, graph racks, right? Really cool stuff, Mitch. If you want a guy, Jason, check it out.

com. Don't miss so much of what we talk about here for the last couple months. We're gonna be really diving and it's gonna be a great show.

Until then, though, there's Alan Shimmel for Textron Gang and Textron tv. Thanks for joining us.