Transcript
Greetings, and welcome to The Pure Report. I am your host, Rob Ludeman. And it is time to bring the orange. And I am super thrilled to introduce everyone, Dan Kent, our new field CTO for Federal, who's joining in the studio. And you've only been here five weeks at Everpure. That's right. Five weeks is the day we turned it to Everpure. I was here. Oh, really? On that same, that synchronicity. It is, it is. Look at that. Look at that synchronicity. It was part of my contract. Yes, yes. Did you have anything to do with the rebranding? Did you know about it coming in? Great to have you here. Thanks for taking, what brought you into headquarters this week? I am here for Geek Council. Awesome. Which is a session where the field folks come in and talk with the product teams and share some feedback and get some views of what's going on with product direction. Yeah, super fun and informative event. I've never actually been into one. I've never been invited. I'm not formally qualified as a geek, but I've always known about it. And what I love about when we do that is a lot of my favorite people come into town. So I'm pleased you got to be a part of that. How did you come to join Everpure? Where were you prior to this and what stood out about what we're doing here? Obviously, I know you have some connections with some of our other leadership, which certainly helps. Yeah. But you do have to make a pretty big decision in your career. Yeah, I'm going to go jump over to this Everpure company and work in the federal space. Yeah, yeah, it's fun. So I've been in the federal space for a long time. But prior to here, I was with a company called CloudFlare. And it was a good company, liked the job. I was an individual contributor going out talking or traveled a lot, talking about technology and how to integrate technology into the business like I do here and like I've done my whole career. And prior to that, I was at a CTO for two different companies, where I actually ran development teams, delivery teams, professional services. And at this point in my career, I said I wanted to run a small team, come to a company that looked interesting to me. My background is in manufacturing companies, so, you know, 15 years at Cisco and Nortel. So I wanted to get back into that and Pure, now Everpure at the time, had a kind of combination. It was a good, obviously, we got a great hardware platform. And then we're getting into AI and data management. And so it was a good fit, right? I could come into a company that was just getting into, you know, driving toward a software plus hardware, really focusing on AI, leading a small team in the federal space. So I thought it was a great opportunity to do something, you know, fun, grow the business with a new team. So that's why I'm here. Timing is really great, I think, given your background coming in, right? Particularly with the not renewed focus or shift in focus, but just the emphasis, right? We've always been around data storage and data management. But you bring in now some of the things that we're doing with automation and with fusion and then the AI layered on top of that. And then this notion of helping enterprises and helping government entities build a true enterprise data cloud within their environment. It's a super exciting time to bring you and your expertise in for that. How did you feel? You've mentioned you're an individual contributor. Yes, sir. For a little bit. Yeah. What was that like? And did you miss the management shops? Yeah. Because that is a different thing. Totally different. If you're managing people, it's one mode. And if you're kind of off on an island doing your own thing, it feels different. It does. It does. And like you said, I've been a leader for a long time, over, believe it or not, 30 years. And that was the first job where I was an individual contributor in about 30 years. So I enjoyed it the first six months. I was learning a lot, going to talk to people. But at the end of the day, I realized what really floats my boat, gets me really excited, is to lead a team, mentor the folks along the way, and strive for the common goal. And so that's why I ended up coming here. I thought the timing was right, to your point. The value of data has become more important than it ever has been. And we're in a great spot because we're, like you said, our legacy in the storage space, data management. Now taking that to really make it AI-ready is not only fascinating, but it's critical for every enterprise and every government agency if they're going to really drive into the AI business. And there's so many of them, right, that have to be worrying about that. And we'll get into that in a little bit. What do you find challenging and interesting about working with federal agencies, right? Because if people aren't steeped in that environment, it might feel like a black box, like we don't know how it operates, but there's contractors and there's so many different aspects versus we're just going to go out to some bank and work with them. How is that? It can be complicated. We have very unique acquisition models, very unique ways of, well, they're so large, first and foremost. So you have bureaucracy and some of the bureaucracy is good. And the way you work with the different partners and influencers within the agencies, it can be complicated. But at the same time, it's got some of those difficult problems in the world to address. It's got some of the biggest data sets, you know, obviously they're supporting, if you look at like Social Security Administration, supporting 300 million citizens. You look at the Department of Defense or Department of War, we call it now, putting systems, you know, compute and storage systems on battleships and putting it on planes and such now. So these are very complicated and great, I would say, interesting and fascinating engineering opportunities, which is why I got there in the first place. I actually, from college, I was in the Army. So I naturally tend to go back into that space. I feel comfortable there. But you know, it is probably some of the most craziest use cases of IT that you will ever see when you look at where we put it, everywhere from obviously the space shuttle to, like I said, on ships and in Humvees, places that you would not think of placing IT components. So we get to do that within the federal space. Yeah, I got some exposure to that when I worked at Sun Microsystems, right, because we had these hardened systems that we would give to a company we work with called Temis, right? And Temis would take them and harden them and then they would show up on a battleship or in a submarine or I think even for a while we had, you know, we had drone systems sitting in the lobby of one of the headquarter buildings because they were taking Spark and Solaris and Throat. So, so massive exposure to that on my side, but also like the complexity. That was when you really saw the complexity of what that is or, you know, there was another time I had Nick Pisaki in here in the podcast and he talked about some of the things that he had to deploy when he was leading IT for the Army and the considerations that you have for temperature ranges and it's really cold here and it's really hot there. Yeah, we go down to Southern Virginia and blow up the hardware to see if it can handle the explosions. Oh yeah, that's really where it happens, right? Absolutely. Absolutely. Wow. So we do all sorts of fun things and then there's some other things like the FAA and very 100% up all the time, you know, very complicated systems or like I said, Social Security Administration, just massive data sets and massive problems that we have to So that's what's exciting about it and I think you're always doing it for a mission, right? And the mission is typically about, you know, citizens. And so you also have that, you know, feels like you're doing something good. It's not just about profit. No, no, no. Are there missions or what are the missions that you've been involved with in the past to use that vernacular that stand out for you or things, maybe something that you're proud of that you played a part in? Yeah, previous company, I think I told you I was at Cisco Systems, we played a big part in the first Gulf War, actually the second Gulf War. And it's when they were really changing the whole infrastructure and how they went to battle with IT, with, you know, TCP, IP, and putting the network and data, right? And how having the, basically the information age going to the battlefield. And so that was just a massive change in how we prosecute battles and you can see it today, right? It's even got more and more so. But you're putting, you know, IP phone systems in Humvees, connecting it to wireless networks and the wireless networks had to actually move with the battalions that were moving to. It just didn't exist anywhere else. So that was probably the most exciting part. But there's so many more. There's so many more. Like I said, and a lot of the work I did at Social Security Administration, which is, you know, it really is meaningful to people. And then at the state and local government agencies, there's just a ton too, because you're now dealing with, you know, caseworkers and how you help them get to see more of their citizens that need help, right? So by bringing IT to the field and getting out to the edge, they can actually be more productive and see more people. And if you can see, they can see two more people a day, that's, you know, it's 10 more people a week that they can help. So it's everything from, you know, that level to a router in space. Efficiency, right? All around, all around that efficiency, but all around really back to what you talked about at the start is, you know, why you're here is really, it's all around the data. And one of the things I wanted to get your perspectives on, because I know you talk about it with other organizations just around AI, right? And so I kind of wrote down in the show notes, like, are we at this federal AI tipping point when we were talking pre-episode? It was, you know, there's a trendy technology or something that comes along every 10 years, 20 years, you know, whether it was networking, going back to when you were heavily involved in that, or cloud 10 or 15 years ago. And I feel like we're at that similar point with AI right now where, you know, the federal government has to make decisions, right? How far in are they going to go? They're going to go. They're going to go. Right. They're going to go. That question from five or six or seven years ago is decided, but how and where are they going to do that? Where is it going to make the most impact? What's your perspective on this tipping point? So I think, well, the tipping point probably was a year and a half ago. Clearly, there's a couple of reasons why. The government probably has the largest data sets in the world, right? And so how do you get information from data? You use AI because humans just can't do that search. So that was one. Number two is the competitive nature of, you know, at least from the government's perspective, what are our competitors doing with AI? And if we know that the technology is so transformative, that if we're not leading there, it can really put us at a massive disadvantage globally. And so there was an executive order put in place, actually, I think, by President Biden put one in and then President Trump actually put in another one. So they're now putting their money where their mouth is. So there's money going into AI. Every agency will be looking at AI. And not that they weren't using AI, honestly, you know, I've been working with AI for at least 10 years myself, looking at the classic AI, you know, statistically, whatever you want to call it. But this new AI with generative AI and looking at agents, how that is transforming the government. They've been taking asset management for at least two and a half years now. There are about 2000 use cases, and that's about a year old. So you can probably assume there's probably 4000 use cases of a government using AI in the different agencies. So every agency will use AI. I would tend to believe that most functions will use AI. So it's not a matter of where or when. It's a matter of who isn't using AI, right? And so it's that pervasive in the government. Most of these are in pilot phase. But they're using everywhere from, like I said, help maintain some of the air platforms that the Air Force uses to battlefield intelligence and what do you do with it to fraud prevention and detection in the IRS. So using AI pretty much everywhere in different places. You know, obviously, the sciences are using it to help with the research. So it's being used everywhere. So the question is, you know, what does that mean and how can we help and why? And it really comes down to AI is only as good as the context, right? We all are very familiar with chat GPT and just using a generative LLM. If you really wanted to help your mission, it's about context. So how do you get your own data, you know, Social Security's data, the Department of the Army's data, the IRS has their own data. How do you then take their very specific data and make it part of that context to the AI so that you can really help move that mission forward? And that's where we come in. Right. Yeah. Which is a common thread, I feel, lately, but also on this program that I have is, you know that you need that context, right? And you need access to the data. But it's really the role of modern infrastructure being able to enable that and not put up barriers or provide hindrances, right? That's correct. I think that's why we're at an interesting place here. How would you characterize then the extent to which federal government has infrastructure in place that you would describe as modern? Like, are they behind? Are they ahead? Or is it really dependent on the different organization or mission, depending on where Yeah. I would say that, obviously, AI, as in the gender of AI, has really only been around for three years. Yeah. Right? So, clearly, I would not be incorrect in saying that the government's data systems and the infrastructure is not ready for gender AI. Sure, sure. Right? So, now, there are certainly pockets. And there are, like I said, most of these new AI tools and assets are in pilot phase. And so, they've been buying equipment for that. But a lot of the infrastructure is not there yet. And so, is it 20%? Is it 25%? Who knows? Right. But there's a long way to go. I think the government has, like I said, more data than pretty much anybody out there. The problem is it's in data silos. Yeah. Right? You typically build a system. The system has its own database. It has its own UI and front end. And you keep that data local to the system. And so, that's great if that's what you want to live in. What AI lets us do is look beyond the single system. So, they have to look at the whole infrastructure and say, how do I now break these data silos so that I can actually glean information from those and help with the mission or what I'm trying to accomplish here? And so, that's what they're writing about now. And there's lots of RFPs. And there are lots of ideas of fixing that. of ideas of fixing that in infrastructure. It starts with the infrastructure. It starts with models that you need to either create or use, and that's what they are now. So I think we're very early on. The pilots are looking great. They don't all succeed, as we all know. Oftentimes when these pilots don't succeed, it's either they didn't really have a great use case that they were trying. They had an idea, but it wasn't very specific, and or the data wasn't accurate. Like commercial and large enterprises, the data is the hardest part of AI in getting your own data in context and getting it usable by these large language models, and that's where the government is, too. Right? So probably worse than others because they've got data that goes back 10, 20, 30 years in some systems that are 10 years old, 15 years old. So and they've been in a constant state of modernization, but they're also the largest IT procurer and user in the world as well. So they just have these massive systems, so it takes a while to get them up to speed. So that's the process they go through. So it's the silos they're contending with, but then also you hit on a great point relative to the data quality, right? It's one thing if they're able to start unifying or where they can or where they're allowed to unify data to let these AI models take advantage of that, but if the data is garbage, right, or you don't know what it is or the classification isn't there, then it further holds things back. Yeah. And I will say that I don't want to paint the brush that all government data is, sorry, is not great. No, no. No, of course. There's clearly some good data. Yeah, yeah. For sure. But it's very, you know, dependent on the application, depending on the process or the system that they're in. It could be, some data might be 20 years old and some data might be five years old. And so you just have this conglomerate of lots of data and lots of systems. So some, and I will say that probably the government has done, at least in the last 10 years, done more to make sure the data is classified better. But it's still in silos and for other issues that they have there. And they have duplications, right, because just like everywhere, you know, somebody has a good data set, you make a copy of it, and then that original data set, the sole source might have changed and the copy of it didn't. So then you have all that concern with multiple copies of data not being the same, some being old and outdated and inaccurate now. I'm living that on a daily basis right now as I am building sessions for our Accelerate conference in June. And I have four different places where I have session data that is being recorded and changes come in and I have to go hit every single one of those. So that is front of mind. How would you say Fed does relative then to performance and scale? Still hurdles there. Like performance is something contextually that I feel like I talked a lot about more on this program and just in generally way at Pure Storage. Now EverPure did, but a lot of it is table stakes. And again, I know there's places where you need that super extreme, you know, that sub, sub, sub millisecond kind of performance or response time. Is performance still an issue? It is. Again, they've got very interesting use cases, right? You look at the national labs working with very high, intense data. Performance is really important. So the national labs and some of the work they do at Department of Defense, some of the work they do for science and research, that performance is absolutely critical. Like HPC types of things, right? Yeah. Okay. Yeah. You know, they're trying to, you know, solve the genome problem. What other big medical issues are they trying to resolve? Or you know, around the environmental protection and what they can do for, you know, looking at the weather system. And all the modeling that they have to do, which is massively compute intensive. They go from everywhere from that to managing, you know, Microsoft 365 services, right? So it's hard to paint the common brush, but they definitely have that. I'd like to say that the federal government is a microcosm of the rest of the company. We've got financial verticals that care about finance. We've got healthcare verticals and research verticals and education verticals. We also have manufacturing and then service providers, right? So they have some very large agencies that act as a service provider for the other agencies. So we have everything that you would typically see in the enterprise. And so, yeah, I think the performance is critical in certain areas, and in those cases, they've been feeding it pretty regularly. So they're pretty good at maintaining. I think in the non-high mission areas, the concern is, you know, that's where the data has, or the money probably hasn't been there to maintain and migrate the continuous modernization of the tools that probably have more concerns around the performance issues. But perhaps it's good enough, right, for right now, right? I mean, there's that thing, like, unless somebody's screaming or somebody's noticing, it's probably good enough, right? That's right. Just to do day-to-day processing. Yep. Shifting to another topic, right, because when we get into AI, you start getting into issues of trust, you get into issues around regulation, and we're having an interesting chat as I double-click it further around ethical AI and bias, and you have some experience in this. Yeah, yeah. How this relates to that. It's interesting. Like I said, I've been doing AI for a long time. In one of my previous companies where I was a CTO, I was working with biometrics and fingerprint matching systems and facial recognition, and this was about six to seven years ago, and that was a big problem in facial recognition systems back in the day. And it was more around, you know, we could, believe it or not, the color of your skin would give you good or bad results of facial recognition. So we've overcome those hurdles for the most part. But as I was telling you, I read an article just two weeks ago on this woman, this grandmother in Tennessee who was arrested for bank fraud committed in North Dakota, and she was arrested because their facial recognition system picked her out, come to find out, they put her in jail while she was going through the investigation. She'd never been to North Dakota. Yeah. So we still have problems, right? AI makes mistakes, and lots of places where AI makes mistakes, we're all familiar with hallucinations and all, but these are, like I say, the mistakes that the government is trying to deal with, has to deal with, and we all have to deal with to make sure that it works well, or what I call self-induced mistakes, meaning if you don't put good data in, you're going to get bad results out, or if you don't take full context. I mean, this was a simple one. I mean, this woman, all they had to do was find out she never went to North Dakota before and they would have said it. Yeah, and they have ways of checking that. Unfortunately, it took them five months to get there. But that's just one example, but there are many others of self-induced mistakes where you just, you didn't update your latest policies and somebody used a chatbot to get something and found, did something wrong because the chatbot gave them an answer that was incorrect. So a lot of data is time-sensitive, and if we're going to put that old, redundant, information into an AI tool, you're going to have wrong answers coming out. So it's that whole idea of making sure your data is accurate, is timely, is critical if you're going to have good AI applications and good AI outcomes. Yeah, worth investing in, right, to make sure that that's something super critical. What about the compliance frameworks? Again, I've covered NIST-oriented types of things in the past, and I know there's some NIST frameworks that have been built around AI, much like other technologies that have come along, where the government has to take a perspective on it and create some standards and some compliance. That's right. NIST has an AI framework. It helps give you a direction. They've also been working within the Cyber Center of Excellence, getting a little bit more detailed on how you deploy AI systems. I'm actually working with them on their Agentic AI security system, so that'll be coming out here. We're probably six or eight months away from that. They open up to anyone who wants to help participate in that. But yeah, there are different groups. This has come so quickly. It's hard for government agencies to move that quickly, right? Their budgets don't allow that. But they have been active very strongly in AI. I'm trying to think the ... They have a program, it's under NCCoE, and the profiles, they call them cyber profiles, and how you should protect your agentic AI. It talks about everything from access controls to if AI is going to act on my behalf, how do my access controls get handed down to the agent, et cetera, how do you prove that and validate that? But there's also other places where you need to track that you're making sure you're doing the right thing when you deploy AI systems. And so I highly recommend that anybody deploying ... And most people now are looking at the governance models that are out there and how they should deploy and back up and prepare their environment for AI. Well, and cyber's always been a huge concern, right? I mean, one thing that makes me feel a little bit more reassured is if there's one organization on the planet that cares a lot about security and cyber security and all that, and then as they're building in these AI models and start working on AI, I feel pretty good because I'd say the federal government has always had a leading position around cyber and cyber resilience initiatives. And companies around the world and governments around the world look toward it, so it's good that they're following that. And like I said, they've got a couple of profiles out recently. They have the AI framework out there. So there are some places you can go right now if you're going to deploy AI. What about ... We'll hit to the third topic that I want to touch, because again, enterprise, commercial, federal, government, any industry is going to contend with this. And you made the point a moment ago, and we even had a customer on that I just published an episode for, things are changing so fast and there's an expectation to adopt rapidly, but the areas around adoption require still humans, right? People to do things. And even our good buddy, if you've met Sean Rosemarin, who loves talking around all things AI, that's one of the common themes that he goes into is, right, this is partly a human problem too. So how would you characterize then the federal workforce, how they're going to have to contend with ... Is it going to be re-skilling or is it going to be hiring in knowledge workers that have this expertise or some combination, and how are they overall going to contend with the skills gap, I guess? I think, and like you said, it's not just the federal government, it's everybody. It truly is everyone. There is a skills gap. I think this has come so quickly. It's interesting because not only has it come quickly and therefore you have that skills gap, but there's also coming quickly and it could be impacting jobs. We've seen this more on enterprises and commercial work. Hey, we're not backfilling certain roles because we know AI is going to do that. That hasn't come yet to the federal government, but it will. And so you've got the, hey, I've got to get better skills, but I also know that if I don't do something quick, my job can be at risk here. And so they feel all the same stress that, I would say, a white-collar worker in any enterprise company is dealing with. So yeah, right now the government does probably better using contractors and bringing in third party than upskilling. Not that they're not upskilling, but it's hard to do that because you probably are aware the government has laid off about 200,000 people in the last year and a half. So a lot of these jobs weren't backfilled and people were now doing multiple jobs. And so they're busy doing their job and how do you get retrain yourself if you're busy doing one and a half jobs? And so that's kind of where the federal government is right now. Not everywhere, but a good part of them. And so they see it coming, but they probably won't be able to reskill themselves in the short term. And the government is using contractors to come in, which isn't unusual, right? The government has pretty much used contractors a lot, especially for modernization type things. So but yeah, I think there's an emotional impact for people and you see it. We talk with some of these customers and some of them know that their job can be replaced. Others might not want to admit it or acknowledge it. But yeah, so it's like I said, I see the same commercial companies, but usually in the government, they're probably more risk adverse, right? So they don't expect to see that now. Like I said, in the federal government with the little 200,000 plus that got laid off, that was a first for government in my lifetime. So they've already got that concern there. So they're going through challenging times. But I personally, being a CTO and being a person that believes in AI, I think this is going to help. So maybe this will help them get to a one person job or a one and a half person job, but they can start using AI to help do the job a little bit more effectively. And so I like to look on the positive side and help them. Well, I like going back to other technologies that have come aboard that have shifted or changed or, you know, the more harsh word is eliminated jobs, but then they've brought about others, right? I mean, I'd go way, way back to everything was pulled by horses and buggies and guys that made wheels and, you know, the carriages and then the automobile came along. Okay. Well, those jobs are all gone. But there's people that need to make tires and there's people that need to work in service shops and, you know, gasoline, all sorts of jobs came about. And so that's maybe where the optimist is. But then you also have skeptics. You say, is there a notion of skepticism around AI and automation in the federal space? Yeah. There is. There is. And I always, I send them to a book. There's this book called Cointelligence by Ethan Mollick. He's a professor at Penn. And he really, the book is old. It's old now. It's 2024, it came out. It's old for AI. It's old for AI. And it was all about how do we, it's not AI that's going to replace you. It's the person that knows AI better than you know AI. And how do you work with AI and treat it as a co-worker so I can be more efficient and so I can be more efficient and effective. And that's where I try to send them to. And effective. And that's where I tie it. I believe this, right? Certainly there are going to be jobs that are no longer needed. But there's going to be jobs that we now will be doing because we are tracking other things and we're doing different other things that we can be doing now. Because we freed up some of our time and resources to do that. So yeah, that's a great book. It helps people that are more on that side of I don't know if I can do this and what am I doing is going to take my job away. It helps them realize that it's not that bad. And you just got to jump on it and start using it. And using it not just as a helper for writing an email or correct your sentences, but there's other things you do in your day-to-day job that it can help you do differently. And so once you get familiar and comfortable doing that, you'll see how it really can make you more effective. Yeah. It's fun to find those though. I'm finding that I was probably more on the skeptic side for the last couple of years. And then you talk to people like Sean Rosemary and others enough. And partly it's that if you don't invest and use this, you're going to fall behind everybody else. But then Sean's whole line in a pod we did a little while back that stuck out was it's like you have a little team of interns. That's right. You have a set of people, digital people, digital functions that are kind of working for you. And so every week that I kind of find a little discovery of, oh, okay, here's something where I can just go invest a little bit of time and maybe automate this task that used to take me a lot more time. And then, oh, I have time to go spend more over here. Yeah. And I think that's where agentic AI, when the tools finally get here to make the individual be able to write their own agents, that's when that's going to happen a lot. And they're going to see that, right? Some companies are already using this, but it's not, I would say it's mainstream yet. Yeah. Because anybody can look at a chat pod and type a question and get a response. But if I want to every morning have my email filtered out and delete everything that I know is junk or respond, let me know what I need to respond to right away because it's looking at my calendar, that's real time savings for me, right? Yeah. Yeah. Now, could I write that agent myself? Probably not. But there are going to be tools that are going to make it so easy. Eventually. Yeah. Absolutely. They're going to make it easy for you to make agents. And AI really is, I think the value is really going to come out, like I said, when it gets beyond IT, when it gets into the function that you do and you're familiar with what your function is, and then you figure out how to integrate AI into the function that you want done. And that could be a case worker going to see children or welfare recipients, or it could be someone inspecting bridges or inspecting gas tanks or whatnot. So there's lots of places where AI is going to be there. You would never think AI is going to be there because you don't have that function. So yeah, that's where it gets exciting to me. And I think that's where we're going to start to see that. And that's where it won't take away jobs. It's going to make someone be able to be more productive in their job. Right? Right. Yeah. No, it's timely you brought that up because I'm having a couple of the gents that you've been hanging out with in the council this week, Rob Quast and Ian Saunders, are going to be coming. I think I have them coming tomorrow and we're talking about vibe coding. And basically, it's coding all around the agentic AI and some of the things that they're doing. And again, we're approaching that ability once the tools become easy enough. Because again, personally, to your point there, I think the last time I coded something was basic in 1984. It's just not something I've ever had to spend time in with my job. But I would like to get into it when it's something that's easy enough that it can help me with some of those functions. Make your own website and put your podcast on it. Yeah, exactly. As we'll close here, and we'll kind of bridge the everything AI that we've been talking about in the Fed space, what gets you most excited that you see we are doing here at Everpure to address some of these challenges and directions that we've gone through with AI? Like what stands out? You've been here five weeks. Five weeks now, right. Some various things had to be beyond the people and the culture, which we have in spades. What about the technology got you super excited? Yeah, to me, and I knew we were going this direction before I came here. That's when it's exciting. Obviously, everyone knows this as a storage company. The data management side of it is really exciting. And so the products we have coming out this year, and we've announced them already, is how we're going to help customers take their data, clean their data, label it, be able to translate or transform it into vector databases so that I can start using that into the context of my AI and doing it through a regular pipeline. So this is no longer going to be a manual situation. This is no longer where you have to physically run a tool every day, two days. It's going to be automated. That's when you start to see how a company like Everpure, with all this data that we help customers manage, really take that to integrate it into these AI systems and continuously update them. So you don't have that problem of an old policy anymore, right? Once that policy has changed, it automatically updates your pipeline, automatically updates the AI tool that's out there. Again, a lot of these self-inflicted wounds that we see, AI mistakes, those will all go away once you get control of your data. It truly is the hardest part of building your AI application. Whether you're creating your own AI models, which we help too, or whether you're using an open source model or a commercial model, but using your own data for context around that and creating, we call it RAG, and deploying a chat bot or deploying some type of customer success tool, we're going to be able to do that and give the customers a way of automating that whole process. And really important, right? You don't have the problems of security when you're making multiple copies and transferring it from one location to another. I have to secure both locations. I don't have the problem of systems that are out of time with each other and it's all going to be source data that you use that will continuously be used in your AI application. So that's exciting to me. I think it should be exciting to most people who are deploying AI systems now because it is one of the harder parts to continuously maintain and update and refresh your data. Yeah, it's an exciting time. And for the customers that are already working with Everpear and working on our products, they understand, but I think they have a leg up right now, right? Because you have that foundation of all the things that we've been able to do non-disruptively and make it easier to manage data. And now you're evolving into Fusion and you're doing MCP models and those are connecting with third parties. And so those customers that are already using our technology get these new layers, right, that you talk about around the data management and the automation and the classification with one touch. And it just, it starts looking like an entirely different picture. You can see the vision of the enterprise data cloud has come to fruition and now we're helping them integrate that into their AI tools that they have. And so it's just a fantastic story. And I think that over the next two or three years, a lot of our customers will be doing just that. Yeah. Well, it's exciting to have someone here like you to take that on because you obviously understand it. You want to do some hot takes with me? Sure. I know you're new. But I try to put everyone that comes on through this. And it's the same kind of three questions, right? I gave you some advanced knowledge. Take a quick sip. But we look at, you know, number one, is there some kind of technology blind spot that organizations can be fed or otherwise are not having? Two, we kind of look at an oops in your career, like thinking back in your careers, there's something that was challenging that you'd be okay with sharing, right? Don't want to put anybody. And then three is the look back, right? Look back to yourself in your younger career. Sure. What did you wish you spent more time or invested in? What do you wish you knew or maybe got certain advice? So we'll go number one, blind spot, like just right out of the gate. What do you think is a main thing that folks are missing? Could be AI related or not, right? I'll stick with the AI. Because I'm in that mindset right now. We're in that mindset. And a lot of this, two things, right? Really, one is if you don't have a solid business case, you don't know what you're trying to do, then you're going to probably fail. And again, when I was CTO of the company before I came here, there was a lot of pressure from the top, from the board of directors to do something AI. And I think a lot of companies have that pressure, not just my company, a lot of companies did. And so we basically started doing AI. And then there was this report that came out last summer from MIT that said 85% of AI projects fail. And why did they fail? It was top-down driven rather than bottoms up, right? And so that's number one. It's not just about the technology. It is about the business and being able to have a true business case. But when it comes down to the AI technology, like I already said, it's the data, right? And I would say that anybody that's had a beyond the, hey, I didn't have a solid business case, I didn't know, I wasn't tight enough with what I was trying to do. It was, I didn't have control of the data, right? And so that would be the blind spot. And that's still out there. But you got to start with the business case though, right? 100%. And actually kind of funny, someone just sent me this cartoon on Slack about an hour and a half before we met up. It's a little simple cartoon, right? And it was a bunch of people standing like this, raising their hands and said, what do we want? And then the other AI, what do we need it for? We don't know, right? I'm butchering it a little bit, but that just happened about 90 minutes ago. I saw that cartoon. It speaks to your business case. 100%. And I think we all have to understand that even if we have a successful AI application, it's still version 1.0, right? So I mean, at the end of the day, we're still in version 1.0 of these applications. So we have a lot of work to get them better. And like I said, the maintenance and making it into a self-evolving pipeline that goes forward, it has to be done. So we have a lot of work to get them solid and strong. So that was number one. But it's going to happen. 100%. Yeah, the value is too much, right? We can see the value across pretty much every vertical in every function. There's so many good things that can come out of it. We have mistakes being made, but we'll get through them like any transformative technology. Part of the progress. That's what goes on with transformation. What's your oops that you're willing to share? Now, there's so many oops I've had in my career because I've been around a long time. I think the one oops I will bring up, and I won't tell you the company, but I was building a product. And this is the naive, I'm an engineer leader, and I know better than you. We kind of go back to the business case. You have stakeholders. And I believed in my project so much so that I didn't feel I needed the stakeholders buy-in. And not that I didn't get it, but I really didn't explain it and fully get compliance, get agreement. And so I went forward with it. And even though it ended up being successful, it was very painful because the, and this is, again, business and technology coming together. It was successful because the technology was so good and what we were trying to build was so important. But it was very difficult to get there because I didn't get enough buy-in from my business partner along the way. And that buy-in meant I didn't explain to him how it could be beneficial to him and why and how this was going to grow his business long-term. And so the lesson learned, of course, is just because you think it's the best thing in the world, if you don't get your stakeholders to believe at least 70% of the way there, it's really hard. You can be successful, but it was so much harder than if he had just been on my side the whole way. And again, I think moving too fast, not really getting the true buy-in along the way, it was important. And again, I think that's important for any engineering person who is working on a project, you need to get the buy-in. Yeah. An old leader that I worked for taught me of that Japanese concept of nemawashi, right, where you go out and it's a word about kind of gaining input from people in the process, right, who are the stakeholders or even extended out from there. And you kind of go out with ideas about what you're doing and you make them feel a part of the process. But it's super critical to your point because then once you did go forward with it, it felt a little hollow perhaps or maybe you didn't quite – It was hard. Because you didn't have that connection, right? You hear this over and over again, put the money, time in up front, it'll go faster in the long run. And I didn't do that. And yeah, this was several years ago. It was a good lesson learned though. But like I said, there's many – I've had many oops as we all have in our career. But that was an important one because I was in a pretty senior role and so I passed that on to my team. The other – yeah, okay, that would be it. That's good. And also, I mean, even as we're later in our careers and in senior roles, there's still instances where we have an opportunity to learn where we can grow and get better or realize that we should have done something differently and that's just part of how it works. Yeah, that's life. That's how it works. That's right. Go back to the younger Dan Kent. Yes. Earlier in your career, what advice do you wish you would have gotten or what do you multiple directions you can go with this one. But really kind of that perspective back to your younger self. Sure. I think it also goes back to the engineer in me, right? So when you're a good engineer and you like doing things, you get good at it. It's similar – well, yes, it's the last story. I wish I had learned how to influence people earlier. Okay, interesting. You know, even as an early – a young leader, you know, when you're the first level leader, you're not really influencing. You're really direct leadership, right? So the older you get, you have to influence people. And what does that actually mean, right? I mean, that's the hard part. It's the softer skills of engineering, right? So I think the hard skills of engineering came pretty good to me, the softer skills. And I think this is fairly common around engineers. It is. So how do you work on those soft skills when you're younger? So how do I influence others? others? When I became a second level leader, that's when I realized influence was more important than just direct leadership. When you're just managing a team of eight or ten, they all know they work for you and, you know, when actually you work for them, right? See, you do work for them. But then when you move up the chain, it's all about influencing. It is get the buy-in. So, in the case that I talked about, my big oops was the buy-in from my peer or my stakeholder. But when you're a leader, you got to get, you know, how do you get influence others around you that you don't have direct leadership over, right? And how do you, you become successful by how you can influence others around you to help you be successful. And that's a hard lesson to learn for engineers, I think. And so, that's the one that the softer skills are really important as an engineer. The, I was fortunate enough to be the cutting edge of many technologies, which was great, you know, early on with the internet, TCP IP, early on with voice over IP and early on with the cloud. So, I was very fortunate to get a lot of those technologies along the way and I would stay on top of them. But the softer skills were a little bit, you know, you have to own that, right? That's because you don't work with them necessarily day to day where you do on the technology side. So, yeah, that's the one thing that, you know, if you want to have a true balanced career as an engineer, and it doesn't matter whether you're a leader or you're building your own, you know, your individual contributor trying to be successful. Influencing others is really important and how do you do that in winning as a team rather than winning as an individual contributor. Yeah. Yeah. I mean, two thoughts on that. I mean, one, there is that transition, that inflection point where you go from that, you're focused more on your direct reports and developing them. And I'm not saying that doesn't stop. But as you gain more responsibility and you have other peer leaders that you need to influence, you have to find that balance, right? You can't spend all of your time, which I found very challenging as I took on more responsibility and a larger team. And I even struggled, right? Because I still was spending a lot of time on the development of my team and what they're doing. And it was a lesson to let go, right? Let them succeed or fail, but provide the support that they need, the resources they need and then get out of the way. But that I really needed to go spend. And I had a mentor manager who actually talked to me about this and said, you are not spending enough time with your equivalents in these other departments. They don't know what you're doing. And so you are missing out on that. Yes, it's great you're developing your team. It impacts your team too. It's not just impacting you because they don't know what your team is doing. You're not helping them, right? When it comes to review period and all that stuff. Yes. The softer skills, like I said, and understanding that, it's really important, especially the further up you go in engineering. Well, and then public speaking, right? I had Naveen, our chief architect from DX, and we did a little JAG similar to what you're talking about with soft skills. He was always an engineer, chip designer, came here 13 years working in digital experience. And we put him on main stage at Accelerate last year and he'd done public speaking. But we talked a little bit about, you know, along with that skill around influencing, everywhere you go, you're going to be doing public speaking, regardless of whether you're an individual contributor or whether you're leading a team. It's just something that you do, whether it's in a small room with a few people, you're going to be presenting, you're representing to your execs, maybe you'd be at a conference, but that's a soft skill. And that's not something that all engineers or many engineers probably get a lot of exposure to in their studies or even early in their career. Yeah, and some of them don't even like it. And so, and that can be limiting. It doesn't have to be. But depends on what they want to do in their career. Yeah. Great retrospective. Thanks for sharing that. Final thoughts? Like, we got to wrap this thing. I got to let you go on to whatever's next. But final thoughts just in general, what you're optimistic about for AI and optimistic about working with the federal space here with Everpeer? Yeah, I think we're going to see, like I said, I'm going to see AI continue to proliferate. Very much looking forward to how we can help our customers in the federal government use it and get control of their data. Now, that's a massive task in the federal government for control of their data. But I think we have a play here and we'll continue to do that. And maybe I'll come back in a year and give you some great examples of what you've been doing. Let's do that. Let's do that. Yeah, I'd love to put some color behind a lot of things we talked about, because this was a good, more than just a baseline. I want to kind of baseline. There's some really great insights. And I appreciate you coming on and sharing. Like, I even learned a ton of things just about how you work in that space. So thanks so much for making the time today. Of course. I appreciate it. Anything you got to plug? Anything you're working on? You've only been here five weeks. Yeah, I'm still trying to get by. Can't expect that you've cranked out a blog or something like that. I am actively working on a blog. It's talking about the economics of data. Okay. But I will absolutely be writing. I've got a ton of blogs out there on AI and agentic AI right now. And hopefully my next one will be about security and AI. Love it. And what if someone wants to, I mean, in your role, you're around to talk to folks, right? Absolutely. Yeah. So if anybody listening or watching, right, you're in the federal space or otherwise, reach out to your peer team. Say, hey, I want to have Dan come in and have a conversation around AI. And certainly I know there's other areas that you would go into. I do a lot with cybersecurity as well. I would imagine. I would imagine. Maybe that's on the next podcast as well. That was great. Hey, thank you so much. I appreciate it, Dan. It's great to meet you. And thanks for being willing to come on after only five weeks here. I love people that love talking and also are subject matter experts and domain experts. And you certainly are that. So I do appreciate it. Thanks. And I'll be back in a year. Let's do it. Let's do it. So thank you, Dan. Thank you, everybody out there for watching, listening to this episode of the Pure Report podcast. So great to have Dan on. And as always, tell a friend, tell a colleague and share it out. And we will keep the great guests like Dan coming on to the program. And with that, we will wrap forever pure for Dan Kent. This is
