The Evolution of Defense Tech with Ben Van Roo

[00:00:00] Ben: I like to see that people want to build in this country and build for this country. But I'm very optimistic that there are real people who are excited to build real things.

[00:00:10] Ben: And I believe over the next 10 years. The government can benefit. I think we need to see more things like trade wins exist. We need to see ways for faster ways for people to get clearances and to get facility clearances and other, just these barriers that kind of slow down progress to continue to be removed or addressed, whether the private sector does or whether the acquisition community does.

[00:00:35] Bonnie: All right, this is Bonnie Evangelista with the chief and artificial intelligence office joined [00:01:00] by Mr. Ben Van Roo, can you introduce yourself and tell us who you are and what your role is? Where do you work?

[00:01:08] Ben: Yeah, thanks for having me, Bonnie. Yeah, so Ben Van Roo, CEO and co-founder of Yurts AI. So a little bit of background about me I've been in and around and working in either the defense tech area or or in artificial intelligence for, oh, God, probably the better part of 20 or 25 years.

[00:01:31] Ben: my whole family's military and for. reasons that I have bad eyes, I ended up geeking out on the mathematics and not joining the Air Force like everyone else when I grew up around the military and the DoD went to grad school, and spent some years working with Rand Corporation.

[00:01:49] Ben: During grad school and after, and then I enjoyed that experience. It's interesting to do policy work. I got to travel around the world and see a lot of stuff and go to a lot of bases and try to [00:02:00] ponder on some longer-term thinking projects. But I wanted to go back into technology at a little bit faster pace in the startup world.

[00:02:07] Ben: And so moved up to San Francisco in 2010. And started working with software companies and what those are focused on different aspects of mathematical modeling. I spent some time working in the education technology space where I helped a company called Chegg grow it and take it public and learned a lot there.

[00:02:29] Ben: But what was interesting around that time is we started playing a lot with the early days of natural language processing, and this is. Oh, circa 2012 to 2013, pretty rudimentary stuff, but it had a lot of potential as we grew those that, that company and the team I decided in 2017 to go back to something really small and I didn't necessarily want to be in a public company all my life.

[00:02:51] Ben: and then I went to a small startup called Primer that was focusing on naturally good processing. So, I did that for a few years, grew a national security business, saw a [00:03:00] lot of artificial intelligence in the opportunities that were both in the enterprise space, in the commercial sector.

[00:03:06] Ben: But what stuck in my mind? We were working with different parts of the department of Defense and in the intelligence community with Walmart and several enterprises. At the time I saw AI evolving language models specifically we're taking a lot of large steps, leaps and bounds.

[00:03:26] Ben: The 2016 through 2020 range and the big problem that kind of remained at our company at that time and just industry-wide is the models were getting good, but everything around the model was still quite hard for enterprises to use. And what I mean by that is, people were, would deploy, let's say an API into a company or they would build something on their own, but it didn't necessarily work with the workflows that they did each day in and day out.

[00:03:54] Ben: It was off to the side and it was good at specific tasks. 

[00:03:58] Ben: But day in day out. What [00:04:00] I saw a few years ago felt like the industry is focusing on the models and not the human experience of how are we going to change what we're doing each day. so what is the value that comes out of AI? And so, what I did was pull a bunch of the people together from my past.

[00:04:15] Ben: Some of the smartest engineers I've worked with and we built a company that focuses on the integration of artificial intelligence into what we do. So, plugging it into preexisting applications, spanning across lots of different data sources at the same time. And what we're trying to do is change and augment the workflows that people have today so that they can be more productive remove some of the rudimentary tasks, et cetera, et cetera.

[00:04:41] Ben: What was interesting is if you look at our business, we have relationships with the department of Defense and different services, and people ask, wow, why did you go to defense tech early? In this area of artificial intelligence, I found that we saw the D. O. D. Leaning in pretty hard [00:05:00] on what we thought where I was going. And it was different groups and specific programs. And there's not obviously, Yeah.

[00:05:07] Ben: There are no programs of records around LLMs or Gen AI as of yet. But people saw this stuff coming. 

[00:05:14] Bonnie: what do you mean? You saw some people leaning in what did that look like for you? they were having more Informal engagements or they were running experiments That or that's what I thought

[00:05:26] Bonnie: of

[00:05:26] Bonnie: what what did it 

[00:05:27] Bonnie: look like?

[00:05:28] Ben: It looked like, more open discussions in various forms through External engagement to trying to run experiments and with, through Kratos with different companies. 

[00:05:40] Ben: this is again, pre-Chat GPT and then the post exploded for lots of good reasons there.

[00:05:47] Ben: But what we saw is the D. O. D. Had a few programs where they invested in artificial intelligence and models. They had examples where they owned the models themselves. They had examples where [00:06:00] companies charge them a lot of money to get access to models. And I think. There was a level of sophistication and knowledge about a world where open source models played a big role and where proprietary models played a big role and they wanted to bring these technologies to bear in a variety of scenarios.

[00:06:19] Ben: So what we saw was an increase in credos, an increase in hackathons increase in just engagement. And we were fortunate enough as a relatively small company to partake in a bunch of those discussions. And, we even have an announcement where one of the contracts we've closed is a BA and it's a pretty sizable one.

[00:06:41] Ben: And so for smaller companies. That seems wow, that was maybe punching above their weight, at the time. But what I would say is there was a need. I think the DoD saw it. There was a mix of companies that were trying to do some things in the space.

[00:06:56] Ben: but it was the right place, right time, I would say, with some of the stuff that we saw. [00:07:00] And then what we're seeing now is. artificial intelligence use cases have exploded. People have put together tons of demos with open source tools, and captains and need threes and need fours just, Hey, I think I can maybe automate this step.

[00:07:18] Ben: And so what's fascinating right now is you're seeing both in the commercial sector and the public sector like a massive groundswell for people to want to bring these types of technologies into their day-to-day.

[00:07:32] Bonnie: Yeah, I think it's interesting how you framed your experience pre and post-chat TVT because I think you're right. There's absolutely a huge difference in being a defense tech company. So give me a little idea of what it's what it was like, especially in 2010. I feel like even 10 years, almost now 15 years since 2010, so much has changed even before the chat.

[00:07:58] Bonnie: What was that like in [00:08:00] I don't know, that era of defense tech versus now, are the challenges the

[00:08:05] Bonnie: same? Are they different?

[00:08:07] Ben: yeah, they're very different. I think what I probably better than I'm close to, I would say is, well, knew the era and this dates me a bit, let's say that 2004 to 2010, but then there was a gap when I was not working with the Department of Defense and working at a different startup, but I'd say, to me the interesting kind of change point in that I like to talk about is like 2016, 2017. So, before that, and it's not that there aren't small companies that aren't working with the Department of Defense and where contracts were won and money was made, but in many ways, the Silicon Valley.

[00:08:46] Ben: as what we would say is assumed in our stereotype that kind of arc of those types of companies over the last seven, eight years has changed drastically. So, 2015, 2016 range, if you are a [00:09:00] company like mine at that time, I think your best pathways were too.

[00:09:05] Ben: Frankly working with In Q Tel was probably your best in my opinion, one of your best opportunities, unless you had a very specific defense background unless you knew who you were selling to right away if you were a tech company and you were trying to work with, like a proper, handful of nerdy kids in San Francisco you were trying to find a pathway, a person, a group that could help bridge.

[00:09:28] Ben: in 18, that was really starting the era of a, and then afterward you saw and well really a and Maven were two big programs where I think The marketing around servers and the intention around open topics, the folks that may have been leaning in and choosing defense or VC back companies, 

[00:09:51] Ben: it was also the low point where this is well documented where Google and was, backed out of working with project Maven. [00:10:00] And if you were in the defense world or if you were in the Silicon Valley world, it wasn't very cool.

[00:10:04] Ben: So, so that was an interesting time. It was the low point, but it really, I felt That was a year that set the stage for a lot of the progress that came. So, from 17 to 20, you saw things like ABMS come online. You saw a lot of progress Palantir was doing well. They had not yet IPO, but it felt like the years where maybe defense wasn't so bad.

[00:10:26] Bonnie: That's a very interesting take on it. I don't think I've heard anyone articulate it quite like you have another element to your 2016, and 2017 time period. The other thing that changed for the contracting community, well, I guess the acquisition community at large was the production follow-up. The feature was added to the NDA.

[00:10:48] Bonnie: I can't remember the year exactly. I think it was 2016, but that was. That was one of the game-changing moments to enable that as a real contracting mechanism or instrument [00:11:00] for a lot of people, both on the government, and the industry side. And I never thought of it, like you said. So of course the things that are happening today make sense because between that and now, to your point.

[00:11:12] Bonnie: There's more familiarity, there's been more time to play, there's been more time to design because these things just don't happen overnight. That's interesting.

[00:11:23] Ben: I was just gonna say that now some of the challenge and the rub that's going to happen in 24 and 25 is depending, on how these companies have projected their finances whatever they assume they may make from the government.

[00:11:38] Ben: In terms of revenue or at least bookings and, single-year, multi-year things type of contracts there might be a lot of pressure, hardware companies, very heavy, capitally intensive. Things software companies, it's less capital intensive, but also like you need ATL. And so, you, I think there are things, again, like a [00:12:00] second front or an Apollo that are part of the process, but.

[00:12:03] Ben: getting a facility clearance, getting the ATO to grow and expand against, you just, in some ways, this is like any VC-backed company, if you said that you're going to accomplish something and didn't really understand this industry and what scale actually means, some people will get bit and I think similarly, you can.

[00:12:23] Ben: Absolutely. Look at the prime contracts that exist and the OTCs and say, probably a really small percentage of those are tied to Silicon Valley-backed companies that may change. But it's not going to change overnight. 

[00:12:35] Bonnie: Do you think and we're speaking in generalities on the defense tech side, startups want to be responsible for scaling, or do they, I also see a side of the industry that just wants to. Build cool new things and so I don't know, from your perspective, do most of the new things that are coming out from startups and whatnot do they want to do it all or do they just want to like kind of niche themselves in the, I just want to,[00:13:00] build and create and do hands-on keyword type of things.

[00:13:06] Ben: I don't know. Cause there's a, there's like in some ways the arc that's rewarded of the startup founder when you're raising money is the 20, four-year-old Harvard or Stanford person that wants to build cool things and they're going to hit massive scale through product-led growth, et cetera, et cetera.

[00:13:24] Ben: And you have to, in your mind, like VCs are not looking for singles and doubles. They're looking for a grand slam on runs that are going to pay off the fund. So in some ways by our very nature, in the early days of a company's inception, people are just trying to address a problem that they think is cool and they want to throw their entire lives at it.

[00:13:43] Ben: And, obsessed about it and get, early gray hairs or whatever. in, in some ways it's the, in the defense tech world, I think it's a little bit different because you can have people at those companies that are super inspirational. But if you look at, [00:14:00] but if you look at Palantir's arc, it was a some very talented, gifted people, well backed, and they went through, a slog of building that business.

[00:14:08] Ben: It was a long time coming, this was not an overnight success. It was successful along the way. It was methodical and it had some real, visionaries, but it had some real talented people who thought about selling and strategy. It's Oh, Andrew, again, like you the cards were stacked and they're all, they've had, they deserve all the success they've had, but.

[00:14:29] Ben: They knew the military, they knew contracts, they had genius product people, and they were extraordinarily well backed out of the gate. And so when you're looking at, a seed investment type of a couple of smart people in their garage, Are they going to navigate the entire DoD process for years?

[00:14:48] Ben: Are they going to be thinking about government relationship strategies? Like maybe not. And so, I think yeah there's real for the acquisition community those are not necessarily dimensions that you can evaluate on every contract. [00:15:00] But is this thing that potentially the DOD is investing in going to be successful or is it, interesting, cool tech?

[00:15:07] Ben: And I don't know, 'cause sometimes that is the right way. Cause you, you end up, hitting a home run with some piece of technology. And so, I don't have, I don't have an easy answer for you there, but I think it happens on both sides in the commercial world and certainly in defense tech.

[00:15:23] Bonnie: Would you say. It is easier to get into defense tech now than it was 10 years ago.

[00:15:29] Ben: Yes, ish. yes, I think that there are more. So, there are more pathways. There are more types of contracts. There is more support from the venture community. And that's even recent. when I started our company. And I said that a decent amount of our business would probably be national security.

[00:15:49] Ben: It wasn't wildly loved. And I think even when we did our series they were mixed reviews on certainly doing a dual use. There's a lot of [00:16:00] like reasons why you shouldn't do that, especially at an early-stage company. But the pathways that I knew existed were much more. What I'd also say is I think in some ways that why the, in the dual, at least my perspective, and I'm a company that strongly believes that we're doing dual-use technology. But I think for the most part, I would recommend you just choose one. Because it is a strong dual-use technology solution. You need to have the technology to make sense in both markets.

[00:16:30] Ben: And a lot of times that's not the case. And then the sales and distribution need to make sense in both markets. Cause doing awareness and distribution and commercial can be drastically different or they both require a ton of energy. And then the last thing is I think the team has to.

[00:16:47] Ben: Get the slog that you're entering. And what I mean by that is, the founders might be writing sibbers and BIAs and, like doing all the documentation and going out to DC [00:17:00] all the time and not necessarily going to where all the rest of the customers might be in the commercial world.

[00:17:04] Ben: And on the other side of it, you look at a technology like ours you can deploy some bare metal and you can put, we are deployed in. Hybrid environments, any of the clouds, and we'll have a hosted solution. And by the way, like we're going to be aisle four through hopefully aisle seven in the not-so-distant future.

[00:17:22] Ben: that's a lot of commitment to infrastructure deployment models. Just acronyms to organize it. And so it's, their advantages it for companies that it can do it, but It's a real challenge to straddle that line. I think for small companies that can commit to going into defense, there are more earlier pathways.

[00:17:48] Ben: And then there is the bridging of the, okay, you've got a SIP or a phase two. There are more things, the tech class stratifies, and the app fits. Some people view that as a kind of sugar. To keep the [00:18:00] company going and find product market fit. And there's not, that's somewhat, I think, accurate, but I think it is also there are needs for those different mechanisms to exist to give companies a chance to make those next few steps.

[00:18:12] Bonnie: Your comments on dual-use are interesting because I'm, not sure I land on either side of the argument, but I've heard this argument so many times and the You mentioned aff-works earlier, too. They highly value dual use as part of their cyber criteria And whatnot, so there's this I would say push or incentive to go dual use.

[00:18:35] Bonnie: So that must be very challenging to navigate

[00:18:38] Ben: yeah, In Q Tel is the same way. they're going to evaluate You In Q Tel in some ways a, the model for a long time was companies that many of the companies that existed that we could try to bridge the gap. Well, Q Tel could help you get to the next phase. It's always DIU, like you exist as a commercial company, and you have cool tech.

[00:18:55] Ben: there's a pathway DAU is much more of a series. A series B series [00:19:00] C, you've made progress as a company and now let's help get you into the defense world. And so there's almost the motions. And I think that there's a lot of infrastructure around that. Cause in some ways it makes sense.

[00:19:11] Ben: But I think there's a lot of really valid arguments that say, what if you're building a drone to do Sure. It could maybe be used. In a civil or commercial world, but like probably not really. And, in software, if you're building an open source product-led growth company, like I, we work with a company called unstructured for one of our contracts.

[00:19:34] Ben: And Brian Raymond's a good friend and a great founder. They're very unique in that they have an open-source play and then plays that are specifically tailored for the government, it's just different, it doesn't happen that often. And so I think the idea of saying, Hey, look, we're going to straddle and monetize.

[00:19:49] Ben: In both markets or both sides of the coin, it's very rare, it can happen. I hope that we continue to make progress in that space on our end. But you have [00:20:00] to have technology that makes sense sales and distribution that can do that slog and then a team that can navigate the sides.

[00:20:06] Ben: And I think that's, challenging.

[00:20:08] Bonnie: Yeah, I would imagine in that scenario, your company almost starts to have double I was going to say double the functionality, but for sales, for example, you need government D. O. D. sales or whatever. And then you need your commercial sales. And so every function of the company has to have two sets of itself to satisfy two different audiences.

[00:20:27] Bonnie: Is that what it looks like?

[00:20:29] Ben: Kind of. I'm going through this right now. We're not a massive organization, but we have people that focus on commercial and we have people that focus on D. O. D. What I would say is that like for what we do, if I'm talking about, Hey, we can connect to SharePoint and Oracle and, any data stores that you have.

[00:20:46] Ben: And we can plug into lots of different applications. On one hand, it's like the tech stack in some ways for JP Morgan doesn't look drastically different than the tech stack for the DOD lots of SharePoint and email or whatever. so, so [00:21:00] for us we can span that. But the language is different, obviously you're going to different events.

[00:21:05] Ben: but maybe to, to pause. If you're in the commercial world and you're going heavily in, let's say, FinTech and you want to do CPG, your CPG salesperson isn't necessarily going to be great at selling to the banks. so on the one hand, I don't want to get too wrapped around, well, is us in defense if you go in the commercial world, they have some of the same challenges.

[00:21:27] Ben: They're a little bit different. you're not worried about zipper, but in some ways, the infrastructure and security protocols that we think about to be able to be deployed on. I'll six is, not probably too far off than what, JP Morgan wants to care about, sorry, we're not worrying JP Morgan.

[00:21:44] Ben: I'm just picking on them today.

[00:21:45] Bonnie: right. Well, what do you think is in store for us over the next 10 years? And

[00:21:51] Bonnie: Defense tech in particular, we started with defense tech and went from 2010 to now post GPT. We have [00:22:00] things more tools in our toolbox. I would say more channels for, barriers to entry or sorry, less barriers to entry, hopefully.

[00:22:06] Bonnie: And what, what's that going to give us the next 10 years from your vantage

[00:22:10] Bonnie: point? 

[00:22:11] Ben: I think there's a belief by some that are going to say the defense tech and VC bubble is going to burst in the near term. And I think that there's a chance that there could be a rough couple of years ahead of us just in part, like The CRs that we have in place and it's an election year.

[00:22:28] Ben: So the assumptions around your immediate monetization in that space are different. I think we will need to fundamentally see some degree of shift in either the amount that the primes are being awarded, or just the percentage of ODCs that are going to some of these startups that are popping up in the world.

[00:22:47] Ben: I do think if you look at some of the startup technologies some capabilities that are maybe traditionally given to a prime and then some subcontractors to snap together Amazon services to do X[00:23:00] some really good COTS offerings are going to pop up that I think will potentially and hopefully naturally erode some of the heavy service area building just tools that exist.

[00:23:13] Ben: So, Palantir was a classic example in that space and what they did, for bad but, and I don't know that it's going to lead to litigation, but I think there are opportunities there. In the midterm, I think we're dealing with a different world. I think defense is going to be a growing industry.

[00:23:29] Ben: It has been, and I think it will be a growing part of the national budget. And I think that there will be more success. So in the short term, there'll be some wins. I think we could see, you could see an Andro exit in an IPO. SpaceX continues to be an incredible company. those are the outliers.

[00:23:50] Ben: I think, there will be hopefully some pretty solid examples of companies that grew up in this generation and this vintage of startups that had successful exits. [00:24:00] But yeah, frankly, like the kind of chip on their shoulder. Mechanical engineering nerds that are in El Segundo are, tweeting how cool it is to be in defense tech.

[00:24:10] Ben: That's amazing to me. I don't, I'm not good at tweeting per se, but I think that I like to see that people want to build in this country and build for this country. And not do paid search ads and there's a place for that. I'm very optimistic that there are real people who are excited to build real things.

[00:24:31] Ben: And I believe over the next 10 years. The government can benefit. I think we need to see more things like trade wins exist. I think we need to see ways for faster ways for people to get clearances and to get facility clearances and other, just these barriers that kind of slow down progress to continue to be removed or addressed, whether the private sector does through, through those companies I've mentioned or whether the acquisition community does.[00:25:00] 

[00:25:00] Ben: So, I'm optimistic. I think we're going to need to see more progress like that, and then let the market, work itself out. And so, it'll be interesting. 

[00:25:12] Bonnie: of tech. And I'm not even going to niche it beyond the generalization of tech. I'm not even going to say AI tech, but just curious if that is something on your mind because, on my side, I sometimes hear commentary from the government buyers, the people who have to consume the tech that there's so much, like even right now, there's so much out there, how do you like wade through?

[00:25:38] Bonnie: All the things curious if that was

[00:25:41] Bonnie: anything has had to any thought and experiments on that, from your side?

[00:25:48] Ben: yeah I don't could go in a rabbit hole on this one. I think that it's exciting to see companies and people have a little bit more of an outsized role in [00:26:00] marketing themselves, their defense tech startups, their defense tech VCs. It's like sometimes, in some ways, it's like when you've been around things like AI and that space for a while, and you see, thought leaders popping up on LinkedIn that have been engaged in this for at least six months, that type of stuff.

[00:26:19] Ben: You're like, okay, you know this? And the same kind of with the defense tech team, it's are we going to, are we going to implode? And there's so much marketing and hype. But I, try to back up a little bit and say. Okay, what's the extreme alternative? And the alternative is that people are not engaged and we're in another generation of people working on things that I don't necessarily think advanced humanity.

[00:26:43] Ben: And so, I hear you. I see a lot of companies that say they do stuff and people that say they do things that are like lightweight wrappers around GPT four, and they're hyping up their solution. And maybe that makes sense. And maybe they're a company, but also Maybe they're just a distribution channel and [00:27:00] it's that's, that puts it, it makes it hard for contracting acquisition officers and I'll speak about AI to say, what's there.

[00:27:07] Ben: How do we assess what this is going to look like in 5 years in production? Is it going to be extremely expensive to support from a computing standpoint? Or is this thoughtful technology? I think the acquisition community has an extraordinarily difficult, you look at like open topics for sievers.

[00:27:22] Ben: That's amazing. But now you have a flood of amazing companies as opposed to, you hear me whine in my past writings about like only a handful of companies getting a lot of that money. You have a flood of companies putting their proposals forward. And we haven't created the infrastructure necessary to truly evaluate this huge aperture opening.

[00:27:44] Ben: If you had a very specific topic, you could have an expert, potentially review, that area and weed through it. Like we're putting a lot more into the acquisition community to be experts in everything. And I think that's hard. I think, potentially like even the liability structure [00:28:00] around, well, did this person get it?

[00:28:01] Ben: The second, how do you decide, how does the contractor, an acquisition officer decide? is it how much they've raised and how many MOUs they've had that also support? Cause I can't sort through all this. And I think that's extremely challenging. if I were to my world, if I were to back up, I'd say, we either have to just neck down and choose X companies and place hard bets.

[00:28:23] Ben: That doesn't necessarily feel like a meritocracy. I think we could say, or we open up the aperture and we make a bunch of little mistakes. And it's just a trade-off of what we want to do as a country and as a nation in fostering the development of companies in the space. I think the only thing I.

[00:28:41] Ben: I think it will be a very good 10 years. I think it might be a bumpy one or two. But I hope is we don't recoil when it's defense startups fail and we don't, companies fail all the time, have some, we all need to have perspective and then say, [00:29:00] did we make some advancements in this space?

[00:29:02] Ben: And that's a longer term. 

[00:29:04] Bonnie: I feel very seen by some of your comments about, the struggles or the challenges that acquisition communities facing and going to continue to face. I would also, think all of your points are very fair in that Where do you want to land with this conversation? What else?

[00:29:21] Bonnie: was left unsaid that you want to share either with industry or the government in terms of like where we want to continue to focus or double down on

[00:29:31] Ben: I think it's a, it's an interesting time to, to just have try to encourage all sides to have a little pause and think a little bit, in, in decades a bit, as opposed to the here and now and the hype of it all. We have a very pressing need in terms of just the global dynamics in the geopolitical environment, that we all live in, we have some seismic changes in technologies.

[00:29:59] Ben: [00:30:00] When you look at AI, you can look at drones anything in space. We have areas of gaps, Everything around EW Counter UAS, like there are some real. Things that can expose the entire industrial military base and the defense world as we know it. And so, like, all right, try to sometimes, even when you're doing your day in day out tasks and hype or annoying things or things you're in the acquisition community and you're just trying to sort out like who's who in the zoo.

[00:30:37] Ben: In some ways, I am trying to say, this is a moment in time that we have to know that it's going to exist and it to, pass in advance. and again, what I'd hate to see. Is a really hard reaction one way or the other, assuming that however, the next, five quarters are going to play out during the election year or whatever, that it's going to [00:31:00] be, it'll be a little bit weird.

[00:31:01] Ben: But I think there's a really good opportunity to make substantial progress.

[00:31:05] Bonnie: awesome. Thank you so much for your giving me a little bit of your time and for Sharing your insights and experience with me

[00:31:12] Ben: thank you, Bonnie. And thank you to everyone at Tradewinds and in the acquisition community. You guys are doing God's work right now. So, greatly appreciate it.