Techstrong Gang – February 20, 2025

Hey everybody. I'm Mike Bazar. Today we've got everything from generative AI adoption rates to things about how we're gonna reimagine the way where you do research for just about everything.

And finally, a little chat about database sprawl. You're watching Textron Gang. We'll be back in a minute.

All right, let me introduce our guest today. We have way out there in the, the wilds of New Mexico. Tracy Reagan is with us again today.

Tracy, how you doing? Hey, twice in the week. I'm doing great.

This is good. We are fortunate how you go. Also, continuing out West, we're with Lisa Martin, who's joining us from the Valley.

Again, she's the CMO advisor for the Future Group and is an expert in all things related to not just marketing, but AI and all kinds of fun stuff. Lisa, welcome to show. Thanks, Mike.

Great to be here. Excited for today's commentary. And finally, we have Guy Courier, who's, uh, also with Futurum Group and is the CTO for the Visible Impact arm, and also an analyst there.

And I must say, just to sheer disclosure for everybody, guy, I think you and I have known each other for 30 years, so if we have some inside jokes between each other, forgive us. Yes, exactly. Exactly.

And, uh, uh, that makes, uh, means we met in grade school, of course. 'cause everybody knows we're in our thirties. Uh, I have to say that every time, um, great to be here, um, in the frigid, uh, central Texas, you know, of Austin, where I am, second day in a row of a very cold day.

As you can see, we don't wear sweaters a lot out here, and really delighted to be on with, uh, both Tracy and Lisa. All right, well, it's 15 degrees here in New York, so It's not much warmer here, man. Which it's just, that's, that's like upside down world.

But anyway, There you go. Well, that, that's a good lead in upside down world. So, um, we have, this survey is up there on Techstrong ai, and it was put together by folks at Orix, and they're a data security company along with some help from Intel.

And it shows that nearly every executive that they talked to is still excited about Gen ai, still putting funding together, still planning on putting these projects together despite some issues. And the issues are not just security. It seems like we don't know exactly where to apply these generative AI tools within processes and workflows.

But Guy, what's your assessment of where are we on this adventure? We went from irrational exuberance, but where are we now? Wasn't really irrational exuberance.

I I kind of feel like it was rational exuberance because the, the possibilities from since November, 2023 were just so, they're just so obvious. And the dangers too, are pretty obvious right from the beginning. Now, obvious amongst us versus amongst everybody.

This is one of these sort of like, you know, uh, there's two lenses that I, I try to take to everything that's going on, because our own lens, doing these shows like this one being in the industry, talking to folks who are implementing and all that stuff, it is really like the top 2% or 1% of people thinking about trying, implementing, and using ai. Um, so, uh, one of the, one of the ideas right now is that, you know, we've, we've, we've left the hype or in the part of the hype cycle now, where people are starting to get down to work and be practical. Um, and I don't think that's true.

I think we're still very high up in the hype cycle, uh, uh, right now. And there's a lot of fomo and there's a lot of exuberance. And, um, I, I don't think most of it is really irrational.

I just, I mean, I love this survey. This was a great survey because it showed all these contradictory things going on. Um, you know, this vast majority of enterprises and organizations pedal to the metal adopting ai.

I will note, as I often say, when these things come up, that 70% or whatever of organizations playing to adopt AI doesn't mean that they're going deep. They may be going really broad and really shallow. So, so the degree of spending and investment going on we know is big, but we just, it's, you know, we we're not getting that from this survey.

At the same time, two thirds of them are more are saying, ah, well, you know, there's security problems here that we're not really being addressed. So, you know, maybe we should be cautious about doing some of this stuff. At the same time that nearly a hundred percent of them are saying, oh, well, I personally use AI all the time.

I'm doing it for all kinds of things for work. Um, so it is a wild, wild west, still the wildest wild west, I think, ever, um, until the next one, of course. Um, so, so, you know, where, where are we at?

That was your question. Where are we at? Um, I think, uh, we're, we're still, um, trying all kinds of, well, so first of all, the, the, the organizations and customers and users are trying all kinds of different things, rolling with every new update and upgrade and just doing stuff.

And, um, meanwhile the vendors are incorporating it all over the place and creating new models and new models of models. Um, like the latest hotness is what the small language models. Um, and for me, my personal favorite latest hotness is AI at the edge.

So it's just all kinds of stuff happening all the time. What was the name of that movie? Everything Everywhere, all At Once.

That's where we're at. You, You know what I, what I think when we're going, what we're going through right now in the, this whole AI discussion and the AO AI innovation that's going on is, um, the word ignoramus. Um, we don't know what we don't know.

So we are looking for the answers. And that's what spurs all science, right? That's what, you know, that's what put, uh, Columbus on a boat to sell around the world.

He didn't, we don't know what we don't know. Science is driven by that, that thought ignoramus. And so it feels to me that, you know, I would've been surprised if the executives would've said, fewer executives would've said, yeah, we're, we're, you know, we're not, we're not doing anything with ai, even if they're just playing with it and exploring it at this point in time.

Um, as I always has, I've always said, if it's gonna make a profit, then we're gonna go do it. We're gonna reinvest our funds to make a profit off of AI if we can figure out a way to do that. So we're trying to figure out a way to do that.

And my complaint continues to be that these, that the tools have not been really driven down far enough to the consumer. We've, that's where money's gonna be made. Not just money may be saved or, or, or processes being, uh, shrunk down, which is all important and it saves money, but how does it make a profit?

And I don't know necessarily if we've figured that out yet. I don't think open AI has made a profit yet. So we have a lot to do.

It's an ignor state right now. So traits, both things can be true, though. Both things can be true.

And I'd love to hear Lisa's perspective on this. 'cause this is really her specialty area. Both of these things can be true.

It's wild West, all kinds of crazy things happening. A lot of people are just going well and doing things, creating things or whatever. And the level of maturity of understanding and of the tool sets and everything is very low, but people are full steam ahead at the same time, like you say, any one of our viewers, anybody who is being just a tiny bit thoughtful about this, can do effective, interesting, productive things.

Even if, especially in an environment where nobody's gonna go and say, what's the ROI? Right now there's nobody's saying what's the ROI really on it because they just see it as a good thing on its own. So, but anyway, I kind of asked Lisa a question.

Yeah. Actually, in the marketing world, it is all about ROII talk to CMOs regularly on my podcast marketing, art, and science. And I've talked to several that are leading AI councils, some of them the second time at different companies.

And so what they are seeing is ROI, with respect to increased conversions, for example, that convert into revenue. So there, there are, they are able to use data science to determine and analytics where their investments in generated AI are actually driving faster conversions through the pipeline. Um, and also being able to get like better messaging into the hands of, of young SDRs, for example, to have better conversations with customers, then go through that funnel faster, get to sales, and then they convert to wins.

So it it, we've been talking about ROI for a long time, and I, I think it would, would be, I would say guy in like the last year, we're starting to see more CMOs that are really dialed into metrics based outcomes, and they're having successes and they're having lots of hand raisers and marketing. I wanna be next in the council to try my idea. So there's a lot of, um, just a strong interest.

Now to your point, one of the points that was made a minute ago, is it, is it deeper? Is it broad? It's probably broad.

The depth level is something that's gonna have to be understood as is. Where else outside of marketing and ops and finance and things is AI being used to understand the overall impact of the organization, but the marketers are leaning into being able to show that ROI Well, that, that's cool. I think I was referring to a different one, which was the question that gets asked when somebody says, I want to try this new and interesting thing that I discovered.

And then the answer is, sure. Explain what the, what you expect the ROI is going to be and what the metrics are, what you're describing. Sounds, um, sounds like the next phase.

So I guess what I'm saying is, uh, uh, I mean, thanks for the correction, surely, but I'm saying like, nobody's gonna say no to, let's try AI for this right now. And, and it's almost like I feel like, you know, the ROI is is, it's a, that's a welcome maturity in how you're using it, but it's not a prerequisite to get the funding. You know, he was having a chat with the folks over at the COO Unisys and he was diving into this very question, and it is at the heart of the thing, it's the depth issue.

And he was pointing out that when they do jump into these experiments, they're trying to apply it to processes that are deterministic in the sense that they gotta be done the same way every time. And gen AI doesn't do that, and everybody struggles with how to kind of insert it into something that is deterministic. So it works great for creative stuff, marketing, I'm gonna create a press release.

It doesn't have to be the same way every time, per se. I can train it to use my language and my nomenclature. But deeper into that, it seems like folks are struggling with how to kind of actually put this into something that will have a consistent outcome because the models themselves weren't designed for that in mind.

So, guy, I don't know, I think, I feel like we're gonna get into a lot of things that there isn't gonna be an ROI and we're gonna pull back because it was, the tool wasn't designed for the job. Yeah. So generated AI is a simulation of, uh, what someone would, what a person would create, generate.

It doesn't have to just be writing, um, based on a, a given context in a given input, the so-called prompt, it's a simulation, it's imitative, it's, you're right, you're right. It's non-deterministic, um, is a terrific imitation. Um, and getting better.

Um, what what what strikes me is this, this maturity level, this increasing maturity, once you get over the, the, the not get over, actually, once you grasp that, you're gonna input a bunch of stuff and you're gonna get an output of a bunch of stuff. Like you say, marketing materials could be marketing materials or whatever it is. And you can apply that to a lot of different areas.

Then you start to realize that there are lots of flavors and ways to do this. It's almost like a framework instead of a tool itself. One of the really interesting things from that interview that, that you had, which I watched, was this idea that, um, I just mentioned small language models a moment ago, right?

Small language models are not an easy replacement for large language models. They just fit better in other, in certain scenarios. Um, and the data element of AI is getting more attention as well, where the quality of the input data, or the style or the type of input data has an important effect on what outputs you get.

Um, so, so there's several factors going on that are showing a greater understanding of, and, and, and maybe a, a, an opportunity to Tracy's point for a little more thoughtfulness in what you're doing. Instead of just saying, grab an AI sho, you know, shove your prompt in there. Take the output, move on with your life.

You're so much more productive. You're producing 22 blogs now in a day instead of only one. Um, because you can, you can, uh, apply generative AI to almost anything to make it work in some way that's better or faster or what have you, but you still need to shape it.

You still need to architect it, you know, along these lines. And we still have to worry about data breaches. This is, you know, was part of that.

The article that we were referred to is that, you know, we, we, we still don't have a way to protect databases from data breaches, much less now understanding if the data breach impacted an LLM or not. Um, so there's a lot of work that we have yet to do in this space. Um, but it's, it a very exciting space.

And, you know, it's, we're teaching open a, you know, a Chachi BT is teaching an entire generation of school kids how to do research in a very, very different way. I would've loved to have it when I was in. Can you imagine?

Oh, being able, it's so much easier. You know, I had encyclopedia, I was in sixth grade. Yeah.

Remember going to the library and making photocopies and spending all your change on photocopies of encyclopedias. Oh my goodness. The amount of the amount of copies I made in college was just ridiculous.

So it is, we are, it is a behavioral change. We are, the culture around data in our relationship with data is changing because of, uh, of generative ai. There is just, there's no, there's no other way around to see it.

And because of that, it's gonna move at a lightning speed, and we are going to make huge mistakes. The data's gonna be wrong. 5%.

How wrong was that? Right, Exactly. Then reality intrudes.

Yes. Mike, you talked about, Mike, you talked about non-deterministic versus deterministic and probabilistic and all that other sort of thing, right? Um, uh, do you think that that, that there is this, because you talk to a lot of these, you know, you know, vendors and, and, and also, you know, their customers and that sort of thing all the time.

Do you think that there's this kind of recognition of that element of it and that maybe, um, I wouldn't, maybe it's other flavors or other ways less, less probabilistic, more deterministic, uh, uh, ways to implement AI that are important, that are a good opportunities right now? Is that the sense you're getting? I think the rank and file in these organizations has a better understanding of that particular issue than the C-level executives do, who don't necessarily know exactly how every process works within their organizations.

So I think, you know, when you ask somebody who's a C-level says, you know, yeah, we're all in and we're adding budget to this thing. Yes. But I think when you go talk to the rank and file about, you know, the person running the customer service desk who's trying to implement this thing, they're probably scratching their heads a little bit about how to actually make that a reality that's consistent.

But there is something in that data point from the survey that I wanted to ask Tracy about that's related to this conversation. If I added up all the people who said that they are building versus buying, I get to about two thirds who say they're gonna be building something that feels like an AI agent. Um, do you, Tracy A, does that sound reasonable to you?

It seems like those things are gonna be a little bit more complex than people realize, and or will people kind of give up? I think though, they will give up, you know, uh, in the early days of my career, I was working for a bank who'd actually try to build a relational database. Very, right?

I was on that team and it was like, why are we doing this? We can just get it from Oracle. Um, so I think that they will try, I think that they will find out that buying is going to be better than building in so many ways that that's the answer with any of these kinds of tools.

But in the beginning, I think that people believe that it'll save money in the long run to do so. So let 'em try. Maybe we'll come up with some interesting, uh, products, right?

We, we never know. I think part of the issue then becomes, and I circle back, this is the same issue we have every time we go buy anything as an application or a SaaS application. None of these things attune to really fit my processes, Lisa.

So do I gotta go back in and customize all these AI agents that I'm gonna get from SAP or Salesforce or whatever, but I don't have access to the way they were trained, so how do I kind of, you know, make that elephant dance the way I want it to. That's a great point. And I think we are gonna see customizations.

I think that's just gonna speaks to human nature. Um, so are things gonna go rogue? I think probably, I love Tracy, your analogy ignoring ignoramus, because I think that's, that's spot on.

I think we're gonna see people want, want to make the agents do what they need it to do without understanding the background from a development standpoint. Will they break things? Probably, um, but ultimately if it helps them go faster, be more productive, stay at cost, I think we're gonna see people, um, in a, in a leading maybe rogue way, stepping in and putting their pedal to the middle Guy would say, I have all these agents from vendors and ones that I customize and ones that I've built myself.

Do you think organizations will have the capability and insights to orchestrate all of that into something that automates the process end to end? And will all these things talk to each other in some way that customers enjoy? Or is that gonna wind up being so hard that we might not see any of this stuff in production for another year?

Two? Well, a year or two would be great, actually. I think that's pretty fast.

Uh, um, of course, you know, then no, no contractor got a contract by saying it'll take me a year or two. They get the contract by saying it'll take three months and then they take a year or two. Um, this is really interesting.

It's a really interesting question because the technology itself provides the tools to accomplish what you just described. I mean, I keep coming back to the, the, the, the shot across the bower, the slap in the face, or whatever you want to call it, the Satya Naela gave us a couple of months ago by say, Hey, you know, SAS doesn't really exist anymore. It's just AI agents now.

Right? Why is that? That is sort of taking to a, a, a, a logical, if, possibly slightly absurd conclusion, this idea that once you start building agents, agents don't require prompts.

They make prompts. They are not truly creative, but they are designed to discover and incorporate without being explicitly told to do so. And so by the same token, the the kind of operational automation, uh, that you're talking about, and the incorporation of AI in a more automated way and allowing things to integrate properly and all the other sort of stuff for most technologies we've had so far that's been long on promise and short on delivery.

Um, which is why we're still using a whole lot of non-cloud and, you know, a whole lot of, uh, uh, you know, less modern systems still. But AI seems to have the, and I mean, generative ai, I don't mean general intelligence coming later. I mean, just the simul ai, generative AI stuff has the ability to do, you know, trial and error and trial and piloting and that sort of stuff at a faster rate.

And still having the human supervision, the operational supervision, and not withstanding the obvious security issues that Tracy brings up routinely and the lack of all those other dangers. You, we could, we could get there, I don't know, in a year or two, but we, we could get there. We could get there because the automation gets automated Potentially.

But just a cautionary tale, we should consider how, what we look like right now when we come to Kubernetes cloud native decoupled environments. We really haven't figured that out quite yet. And now we're gonna be adding, um, agents, and agents will have versions.

Uh, this was, this is not an easy, uh, configuration to manage overall. So while we might be producing these products, we haven't really gotten better at managing these massive decoupled environments where we have agents running in lots of places. I complain about agents all the time because I understand the, the complexities of them.

And we, we, Oh, just standard agents, right? Not AI agents. Those are platform, Just regular agents.

Yeah, regular agents. Just what we're managing now, right now, let's add the, you know, the, the, uh, the AI agents to the, the puzzle. We are c creating lots of dependencies, lot, which the death star is growing and growing and growing.

And that is possibly why the platform engineering is getting more and more attention because it's becoming more and more complex. And while we sit there and we talk about a, you know, generative AI and, you know, LLMs and AI agents, we aren't really talking about how it really impacts our production environments and supporting it, maintaining it, tracking security, IT issues and data breaches. That conversation has to happen at the same time.

But guess what? It's not often profitable. So we don't spend money on it until we absolutely have to.

And that is going to be the downfall of a lot of this, uh, this new technology is that we're not able to keep up. Here's my closing prediction on this conversation. I think a lot of organizations are gonna discover that when they apply AI to their processes, the processes have more exceptions than rules.

And therefore they're gonna be like, well, what do we do if this, then that and the other thing? And it's just not gonna scale the way we hope. We'll see how it all turns out, but doesn't mean we shouldn't keep doing this stuff.

But I think we, we need to go in with our eyes wide open. Anyway, we'll be back in a minute. Discover Textron Group, the epicenter of tech innovation.

We are your go-to for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more.

Join our satisfied clients, let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Textron Group. Alright, folks, we're back and we're gonna talk a little bit more about ai, but this time in the context of well researching stuff, and we all do it every day.

We go shopping online, we look around for stuff, we basically hack our way through search results and occasionally call a friend. But, um, now we see Perplexity is the latest gen AI company to join this race, to create these research tools for people where I basically can now log into these tools, type in what I need, and it will come back with all kinds of fun and interesting stuff that is better than a search engine. Lisa, is this changing the way we're gonna shop in this world and on our whole online experience?

Oh, I think so, definitely. I look at it though, when I read it through an academic lens, what Tracy mentioned in a block. I just got these visions and I'm going up to my alma mater, San Francisco state today to meet with the Dean of life sciences to talk about tech and science.

It's where I got my masters, and it reminded me of all the hours I spent in the library. This is in the late nineties, early two thousands photocopying and thought, what if I had a tool like this to help me with my research? How much faster would I have gotten through the program?

So kind of looked at that from that perspective. But you're right, Mike, it is gonna change shopping. So perplexity, deep research AI tool was launched on Valentine's Day.

What a romantic thing to give to your Valentine, uh, in a freemium version. But what it's doing is performing dozens of searches, reading hundreds of sources and reasoning through this material to deliver a, a comprehensive report. It's, I also read that it's, it's accuracy is about 21%, whereas chat GT is about 26%.

But I see some huge benefits here from a time saving standpoint. Um, you know, it's built for people right now who are doing like really intensive work in fields like finance and science and policy. And that's why I brought up the whole science reference there, um, that need precise, reliable research, but also discerning shoppers looking for what we all have an expectation for is I'm gonna get this hyper personalized experience with really relevant recommendations that typically require a lot of deep research, like on cars or appliances, things like that, that require us to go well beyond a casual search.

So I do think we are gonna see a changing the overall shopping experience for especially things like cars and appliances, but also other things as well. I also think it's gonna change academia, which I, like I said, I just, boy, I can't imagine what would've happened if I'd had anything like this back 20 years ago when I was doing my master's degree. So it's an exciting time.

And there's also a lot of benefits in it for marketers that we can unpack too, Right? We just Got The, I think the interesting thing about the perplexity is, uh, is that it's actually taking the next step in creating a PDF for me, which I actually love that idea. If I get something that's right, I want to, uh, create a final document.

And I do think this is something for researchers. This is a very academic in terms of shopping, let's just chat about that for a minute until they can figure out emotional how to, to kind of, um, connect to the, the emotion of a shopper. I do not think research makes that big of a difference.

Now that being said, I know consumer report shoppers, they wanna know every single bit of it, and they will buy the car even if it is dog ugly because, you know, consumer report or you know, a driver magazine sits the top car of the year because that's how they shop. But that's not the majority of people, and that's not how we are bringing up, uh, the new, uh, age of shoppers, which is a TikTok shopper. Great point.

So how is AI going to, you know, start pulling in that kind of information? Because this content generator convinced, you know, all the young ladies to wear a particular kind of lipstick because they look so good on TikTok, and that's how most shopping is done. So we have a long way to go before we really can marry between a consumer's emotion, you know, and, and real data.

Because we do not shop, we do not even elect based on data anymore. So there's a, there, there's a big gap between the reality of data and how we just make decisions. Can I challenge you on that a little bit?

Sure. I would love for you to tell Me more better, just, just, I, I I kind of feel like you're right, but as a, uh, as a, as a good AI commentator and analyst, I have to say like that's context it context is, is, uh, maybe the most critical component of an AI prompt. So if you have a system that's not really using context, that's one thing, but if it can use context that is personalized to you, um, doesn't that at least partially address the, the don't Underestimate the influencer?

Yeah, I don't know influencer, I'm gonna go, I'm gonna go back in time and remember a computer shopper and you used to kind of go looking for gear. Yes. Looking for gear.

Yeah. I worked there in, in grade School and looking for stuff, and you know, it, I'll be honest, it was with a heavy sigh that I opened up computer shopper because it was kinda like, okay, now I gotta read through like all this stuff in here to figure out which of these devices and PCs might be the one that I'm gonna want best. And there's all these kind of recommendations, but there's more recommendations that are conflict with each other.

So it was, it was difficult, but guy, you were there for that time. And if I still look out on the web, there's still all this, you know, buying advice content stuff out there that those websites just go away and they become like endless data services for these kind of research goals. It's so touching a nerve with me.

This is where I get so confused. Okay. You know the story like, uh, you know, the, the, the acolyte ask the, the guru, like, you know, what is the world based on?

And the guru says, it's on the back of an elephant. And then what, what, what is the elephant standing on that's on the back of a, I don't know, a duck, I don't remember, sorry. But ultimately the guru says, you know, that, that that animal's on the, on standing on a turtle, and then the acolyte says, well, what is the turtle stand on?

And after a pause, the guru says, well, it's turtles all the way down at that point, you know, what are we ba So this is, I get, this is where I get confused. Okay, so where does this end up? The AI are being trained on, uh, online data that is commentary from, normally from people on other things that were ultimately created by people that eventually became digitizations or commentary on things that were offline, like computer shopper.

Great. We still need that stuff, right? AI can't create original stuff.

It's a simulation even, you know, artificial general intelligence, which is a GI, which is coming, and supposedly that'll be able to think and create original stuff maybe at that point. But by that point, it's what this I I'm my mind melts it is, it is a re it is a self perpetuating machine of 'cause then you have sim what's it called? Uh, synthetic data to help train ai.

Synthetic data isn't real data. Like, is this just, everything's gonna turn into tropes and cliches and, and we're just not gonna notice every recommendation is gonna be, it's like the Waze problem. I if Waze routes everybody around the traffic.

Well now there's traffic where Waze routed everybody to I, I'm a little, that's, I'm, I'm babbling at this point. I, you know what I'm talking about There. There's never been time even in software where the, let's just use a software, for example, where the best tool actually wins.

You know, A Oh my God, Tracy, why are you saying that? That's exactly what I've been thinking about for like three months. This is, this is, I gotta talk to you later because I've Been, al pointed it out to us a couple of shows ago.

He, when I talked about the democratization of, of venture capital, and he said, it doesn't work that way. And I said, I know. And that's part of the problem because we will go for the top three and that's where all the money goes.

And once those top three have been defined, they don't have to be the best tool. They just have to be the one that most people are investing in. And that's the tool that we get served.

This is part of the problem with everything that we're doing in our culture right now. Right. It is a, a, um, it's, it's a, it's a, it's a mass, you know, run to the, to what we believe to be the top, even though it's not.

That's why I'm saying AI and shopping is gonna be difficult because there's so much emotion and so much that doesn't make sense that we end up getting served the top three tools that may not be the best. So let me pause at this to Lisa here. 'cause to guy's point about, you know, how the models get trained.

Imagine a world that, that goes like this though, right? So instead of the AI model looking for some website, uh, hoover up the data from what if we just, you know, punch the data directly into the model from the people who created it in the first place, and then train the model to analyze the data to come up with the recommendations. But couldn't we skip a step here, which is, you know, I don't need the interim boatload of data into a website like computer shopper just saying, I dunno, that's a good question.

I wonder if, if you, we did it the dumping the data into the LLM route, would that be more bias? Would that produce more bias rather than being able to go out and, and, and trove the web for data that way? I I do really wanna, um, point out what Teresa said, the emotional part.

I'm not a TikTok shopper. I am an Instagram shopper, which probably is just as bad. But you bring up a great point about the emotion piece because a lot of people make decisions emotionally and not on data, depending on what they're buying.

But I think ultimately, um, the, the research tool, being able to go out and pull data from the web, I think, I think it might be less bias. I'm not sure as from a, not being a technologist, but I, I just, the bias thing comes up in my mind. Well, I mean, to Tracy's point, if, uh, things worked out as they should based on capabilities, we should all be running OS two on our risk platform, right?

About an hour, right? Yeah, Exactly. And, and, you know, and, and, and yeah, and, and, and watching training videos on our Betamax.

Yeah, I don't know about that, but OS two should have been king. I, I'm gonna hold to to that point And risk. I'm a huge risk fan.

Oh, Risk was awesome. I know. Well, it what happens, Quin?

I mean, arm is, arm is risk. It's arguably, you know, like there's a lot to be said for risks. So, All right, well, I have to say, I spend some time in the channel, so I know why the best thing doesn't always win, but it has nothing to do exactly with technology as much as it has to do with routes to market and who made what available when.

But we'll come back to that on another day. We're gonna come back in a minute and chat about what's going on with databases. Stay tuned.

com is the leading resource for news analysis and education on challenges facing the cybersecurity industry. com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more. com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more.

com to learn more. com. Home of security bloggers network.

All right, if we haven't made your head spin, we are about to now, Oh, my head's still spinning, made my head spin. There is a thing happening in the world of databases, and every time we turn around, there's a new data type. And this is not a new phenomenon, but there's a lot more of 'em lately and starter most recently with maybe all this vector stuff and graph databases.

And the challenge then becomes, do I need a database for each type of these things? Or do I do these multimodal databases? And there's arguments for both sides of this thing.

But Tracy, let's start with, you're closest to this thing. Do we have too many database types and can we consolidate these things or do we need 'em all? Well, the conservative side of me says we have too many, um, the more, um, curious side of me says, we don't have enough.

Uh, you know, the, we have been struggling with the data deluge for forever. This is, this is never, this conversation has never stopped, um, from the beginning when we were trying to figure out how to search a flat file faster to moving into relational databases, to, you know, uh, graph databases. We are always struggling with data.

And I really believe that the AI conversation has caused that to be more important, how to manage data. Uh, but I do like the idea of having, you know, different databases based on the life cycle is a bad idea. Let's just put put it out there.

I have never supported having, you know, a, you know, a Postgres database being used in development and then moving everything over to an Oracle database in test and production. That is just a bad lifecycle management process. But the idea of having different databases for different domains and storing data in different ways as it needs to be stored as a, as a, that is appropriate, is something that we really should be looking at in a, in a deeper way.

But as I talked about in our earlier session, we have this issue with these very complex configurations and how to manage them across the lifecycle. When we think about technology, we often stop thinking, we think about the cool part of the technology, but we don't think about the implementation and the ongoing maintenance and the sustainability of it. So the more we have these very different databases, the more we need to get that, we need to get better at that whole conversation around platform engineering.

And the problem is, DBAs have never been part of that story. They have sat in their own side of the house and wanted to control the data. And I, I have seen companies that wouldn't even let a developer, uh, you know, uh, check in a, a sequel statement for goodness sakes, to make a single change on a table.

So there's a lot in our culture that we will have to adapt if we are going to improve on how we are managing data. So conservatively, I think is a bad idea from a pure technology growth. It's time that we have these multimodal databases.

Um, it's time. We have different types of databases depending on what we're storing. And it's time for the DBAs to get, become part of the lifecycle so that we can start managing it.

Because if we can't manage it, we are, it's going to be hell Guy. Every time I turn around, somebody is complaining about the total cost of it. And I almost invariably trace it back to all these databases that we're supporting and all these different data types.

So, um, you know, are we kind of our own worst enemies? We absolutely are our own worst enemies because we are tinkerers, we are engineers at at least, you know, in spirit. Uh, we got into this because we like systems and how they work and directing them and organizing them and all that sort of thing.

Um, I think another one of these perpetual conversations is, uh, business value of it, so to speak, like connecting those things, uh, making statements about the systems and, and, you know, investments and all that sort of stuff that, that is, um, not a technical benefit of it. If it happened that it happens faster, but the, the business or organizational benefit of we can enter new markets, we can serve larger constituencies. And so, um, there's this gravity, we were actually talking about it earlier.

There's this real DIY gravity, and I think to some degree that's this, multiple databases versus single databases discussion. The, the use of single database. Um, I, I have a lot of affection for that sort of thing, that sort of approach, because it's simple and, and, uh, so love, as you could tell Tracy, talking about lifecycle and operations, because so many times problems are engendered early in development that only get, you know, seen and addressed, you know, in ops.

And so what I was saying was that single database, single, single, uh, system, um, can be a huge, uh, uh, uh, si time and money saver because it shifts a lot of the burden on ops and it, it, it shifts the shifts the burden of service resiliency and of efficiency and of continuity and all those things. It it more in one place than in multiple places. Um, but yeah, Mike, you know, you, you, you're right.

Um, it, we don't want to take the creativity out of it by any means. And, uh, flexibility and freedom of choice all through the whole application chain, uh, engenders that and enables it, but it also builds cost and it builds risk. Mm-hmm.

So, I don't know, is that even solvable? I'm not sure, Lisa, I've said this before. When we have these very IT oriented issues, it's really your fault.

And let me explain why. Okay? Please do.

Poor marketing. So All these different departments that we have go out and hire developers, and then they go build some application for the marketers who are enamored with some new data type, whether it's video or whatever it may be, or now it's Vector for LLM without ever considering the fact that none of this stuff aligns with their existing IT investments. And then they go build this lovely app and they play with it and run it for about six months before they get tired of it and hand it off to an IT ops team that Tracy described that already has a bunch of databases.

And two things happen. Either one, they just suck it up and manage that database, or b, they try to convert it into something that they were already running with mixed results. This is at the heart of the divide between business and IT is there, can we do something about this, Have to be done.

It's in it's collaboration from, from the C-suite down businesses, bus lines of business have to be aligned with it. They have to understand what each other's doing, because to your point, the sprawl that you just talked about, it becomes a huge problem, becomes a cost, uh, to organizations. And it ultimately can result in inefficiency and impacted productivity, which is the complete opposite of why, say the marketing folks decided to spin up these apps in the first place.

So I really think it speaks to an ELT being highly collaborative and, and the lines of business, not just marketing, but ops, finance, others, et cetera, sales needs to be in lockstep with it. This is what we need to do, this is what we want to do to deliver this. Is this the right thing?

What do we already have in our environment that might help already solve this problem in a better, more, um, integrated way? But so we ultimately think it's behavioral, it's it's people, it it's collaboration absolutely needed. It's always is Behavioral emotion, right?

There's always, there's always what I like to call an internal Bill Gates at every company. And it's probably a developer who's worked really hard, written some really interesting code, and that internal Bill Gates gets to go and he will pitch something to his director who will get very excited and pitch it to the CTO. Now the CTO is excited about it and they go ahead and they move forward with it without bringing in everybody it, and it's just that it only takes one kind of hot shot to make that decision for the whole company.

And that's how we do business. That is how software, that's how we move forward. That is always been the case and always will be the case.

But on another topic in this area, there is, you know, there's different kinds of databases, but there's also different kinds of data. The more that we can start looking at our data and how we write software, even if when it comes to microservices, the more we can start defining domains and building building systems based on domain-driven design, the better we will get at deciding what type of database should be used, where we wanna store the data. And even if it's the same database that we're using across the whole organization for just basic data, I totally believe in distributed databases.

I believe that we should break up the data because I believe that that will help manage data breaches because they don't have as deeper reach. Once you, you, you, you condense the information. Even if we had the addresses, right, the address and phone numbers separate from the name and it was connected based on an ID that pre prevents security breaches that are gonna impact our identity.

So we don't think of that because as I said in the earlier one, it doesn't, it's not necessarily profitable, but it makes sense for the, in the, for the consumer to start thinking of how we can build data around a, a domain model. And if we do that, then we are able to say, well, this domain has this kind of data and it may need a different kind of database. So just to have a, just to have this, this multimodal, um, database environment, it only makes sense if we can figure out the why we need to do it.

And sometimes the, the internal Bill Gates guy, he's not thinking about the why he is thinking about, it's really cool to take apart your mother's stove. I've done it before and it doesn't work out too well. Yeah.

And sometimes that's super exciting idea. It turns out that there's three commercial products for it and two of them, or, I mean, I'm, you know, I'm, maybe you're not talking about that case, but you know, there's lots of ways to what, what's the, uh, Tim Tote? That's the, uh, that's the pearl expression.

There's more than one way to do it, right? Tracy, you made this point and I just wanna bring it home a little bit. So how come the DBAs don't wanna join the DevOps club, man?

Is there, are they just too cool for school or what's going on there? I have never understood that. Is it a, is it power data is has power.

Um, they don't want other, you know, okay, if we go back to the culture of DBAs, there was a point in time that DBAs were the highest people paid in the organization by far. Um, I can remember, 'cause I was a consultant starting in about 27. I hated doing, working on databases, but my girlfriend was making almost three times as much as me.

'cause she was a DBA and I'm coding, right? Um, and so there's power and money and the culture, I believe started that way and it's never changed. Um, the DBAs have always been separate from the rest of the, uh, development organization.

Uh, you know, deploy hub, uh, part of our product was to help manage database, uh, configurations as part of a deployment. DBAs didn't wanna do that. They wanted to have a, they wanted someone to send an email and say, this is what we need done, and they wanted to do it themselves.

I don't know why that stayed that way. I would love to know the answer to that, but I really believe it starts from the very beginning when d when, uh, d relational databases hit the market, there were really high paid DBAs and they were set as a class of citizens within the culture. Very different from the development team.

They owned it. It's Not, it's isn't it because, or is it possible that it's because, um, that that kind of a role attracts someone who is exacting and controlling and controlling in a, controlling is the wrong word that has such a negative connotation. But just someone who, uh, who crosses t's dots i's checks, grammar, checks, everything like very uh, detail and quality oriented, which in some ways, like the move fast and break things idea sounds anathema to me to a database or DBA mindset where, where quality, continuity, and controller extremely important.

And DevOps in a lot of ways is a way of saying like, you know, uh, you know, let's, let's get things done quickly and fix them later such that they need to be fixed. Let's have something minimally viable. We know that there's a lot of improvement to be made.

That's a mindset and a culture that seems really different to me. Yeah, guy, I think the word you're looking for, it begins with an A ends with an L and has four letters and you can figure out the last from there. Um, I'm going to kind of end this conversation here, but I would just point out one thing.

We in it, we all seem to take pride in how rational we are, but once you get past that first laier, That's the biggest trope. Come on. We're as irrational as anybody else there is out there.

Hey, if Not more. So There you go. Hey guys, thanks for being on the show today.

You guys were awesome. I wanna thank everybody for watching this show. You can find more and a follow on text drawing tv.

We invite you to check all that stuff out 'cause it's just another great day of content for us. Take care. We'll see you next time.