Human-Centric Autonomy for Defense with Scott Sanders

[00:00:00] Scott Sanders: there's really no reason to be losing lives when we have the capability not to.

And autonomy is not going to solve every problem today. And I would guess in the future it doesn't solve every problem either. You're going to have humans on the loop. The question is, is it one to one? Is it one to four? Is it one to a hundred? The only way to figure that out is to get systems out today and have people actually use them for a sustained period of time versus just limited scale soldier touch points. 

[00:00:24] Ryan Connell: Hey, my name is Ryan Connell and I'm with the Chief Digital Artificial Intelligence Office. Joined here today with Scott Sanders. Scott, how you doing?

[00:00:49] Scott Sanders: Great. Thanks for having me.

[00:00:50] Ryan Connell: Yeah, absolutely. I'll just kick it over. You give a quick self introduction.

[00:00:55] Scott Sanders: Sure. so I'm the chief growth officer at Fort Terra. We build full stack and ground autonomy for both defense and commercial applications, especially in the industrial space. prior to that, I had a brief stint at Vannevar Labs and before then was an early employee at Anduril building out the counter UAS and air mission defense, business.

And part of that had a quick detour in life at the United States Marine Corps as an intelligence officer and special operator.

[00:01:19] Ryan Connell: Awesome. Hey, appreciate your service. I got a book that I use pretty regularly. It says,teach me cloud, like I'm 10 years old or something like that. can you teach me autonomy? Like I'm 10 years old. Like what is full end to end autonomy? Like, what does that mean?

[00:01:30] Scott Sanders: the analogy that I used for my mother, who's probably going to listen to this, is we do like way more for cruise missiles, right? Like I want to get a vehicle from point A to point B without a driver in it and it has to go through all kinds of complicated terrain. It might have to operate just like a robo taxi where it's going to go through terrain that's been mapped out.

That's and high definition maps have been created to get you from point A to point B, but it might also have to go through areas that it's never seen before. the old school Roombas that somehow found their way around your living room, but with a little bit more intelligence behind them so that you're not just kind of wandering around, bumping into things.

but the, the way we look at it is this has to be human centric autonomy. This has to be a way to enable the warfighter to quickly and rapidly in a stressful situation get those vehicles and those platforms to do logistics missions or weapon systems missions, or in the commercial space, be able to move things inside of a port or, Or yard without having to have someone physically sit in a vehicle for an extended period of time.

so it has to enable those people to do their mission. So that involves building the front end back end, the on vehicle stack, the hardware to make that happen. that can be highly reliable and function. without access to a network if it doesn't have one and it has to be field serviceable by 19 year old kids who we send out to do these incredibly difficult and challenging missions.

so it can't be a science project. It can't be a bunch of duct tape and super glue and servers in a rack. This is kind of just stuck onto a vehicle. It's gotta be a very tightly integrated platform. and that's, that's hard.because the things you experience in defense are very different than what you see in the standard sort of commercial robo taxi space. robo taxi operators don't have to deal with heavy dust and snow.and a bomb crater that wasn't there the last time that the vehicle mapped it. So, you know, we optimize on how do we enable naval operators and soldiers to, to go do these missions.

They've gotten better.

[00:03:27] Ryan Connell: no, it makes sense. And when you said Roomba, my immediate thought was how, I think I have like the original one that somehow still works. but by works, I mean, yeah, it bounces into every cabinet and wall I have, it, it's pretty good about not jumping off the stairs.but yeah,

[00:03:44] Scott Sanders: their, in the room of technology, department. But yeah, like those first ones were less autonomy, more bump into things and figure it

[00:03:51] Ryan Connell: sure. And yeah. How do you, with obviously without any sort of like IP conversation, but like, how do you get to a point where you're comfortable. And is it simulation based or like comfortable with, Hey,I feel good about this product being out there in a space it's never seen before.

and I have whatever percentage confidence that it's not going to just hit the dirt or hit something that shouldn't hit, like what's that safe space look like for you? Yeah,

[00:04:17] Scott Sanders: of things. one, obviously assimilation is super critical to making sure that, you know, especially when you're pushing out development builds to a test site, you know, it's going to work, the first time and it allows you to do, significantly more traceability into your development process and what works and what doesn't, because these are extremely complicated systems of systems, right?

And you're integrating onto a vehicle that you probably don't own. we don't make ground vehicles. I envy the Tesla folks, who get to, the drive by wire guys in theory sit next to the FSD guys, or at least on the same, same network of communication. but, we look at this as a, how do you do requirements traceability on the engineering development side that then traces into, Everything from the test requirements down to the test cars kind of as you come down the left side of the engineering V.

and then that thing has to go into hardware in the loop testing. So know, your bench setups of the components to make sure that it builds and it runs. And you can then run hardware in the loop testing. And then you get into software in the loop tests, or go from software in the loop to hardware in the loop.

And then on to, vehicle in the loop testing, which for us is, at a test range that we operate. Almost every day of the week, sometimes, and often for 10 to 15 hours a day, depending on what we're doing. Because you just have to get the physical miles on these systems, We work in a space where you can use generative AI for a lot of stuff, but you really can't do it for path planning.

And behaviors, because the behaviors you're asking these vehicles to do are the first time someone has seen this exact scenario. Unlike on a road where, There's only really one way to do a right turn at a stoplight. You either can make one or you can't. it, we deal with terrain that's changing all the time.

And every time we operate a test range in a place where it snows six months out of the year.whatever we learned about snow when we were kids about every snowflake being different, it's definitely true. You get different types of snow and you can't just use, okay, we collected all this data and this is what we think the vehicle should do in this space.

There is a lot more puzzle solving than pattern matching. You can do both, but you really need to focus on The puzzle solving and it has to be able to get through safety and evaluation. So you have to be able to ensure that given a certain set of inputs, you're going to get the same series of outputs on the backside.

so for us, it's doing all of that, in partnership with our customers, to enable the deployment of these systems today and not in 10 years or 15 years. And really focus on how that operator then gets that end product because the product isn't the code, right? The product is an end to end capability.

We always make a joke that I could put a purple pterodactyl in the driver's seat and charge a subscription license for it as long as that thing went from point A to point B. No one really cares there's a pterodactyl there. That might not be the case everywhere. CDAO where you probably do care about the code.

And that's obviously been a little bit flippant to the software engineers, but the end state's kind of the same, like you have to deliver a relevant product at the day.

[00:07:09] Ryan Connell: no, that makes sense. So What's so intriguing about this and what you've said so far to me is, I typically, like you just pointed out, I live in a software world these days. it seems like, is it equal? Like, like what's your day to day? Like, is it hardware or software? Is it 50 50? Is it mostly software, mostly hardware, or is that a hard question to answer?

[00:07:26] Scott Sanders: I love, my favorite military style answer. It's a, it's depends, it's METTC dependent. and for all the former, DoD personnel listening, like METTC, I forgot about that useless acronym. You know, we think that you have to, there's a couple of things to make autonomy work in this space.

And one of them has to be priced at a price that the DOD and our industrial partners can buy it. And if you're going to go put,tens of thousands of teraflops of compute onto a vehicle, it's going to cost you a million dollars. hate to break it to everyone, the DoD is not interested in that solution.

and neither are your industrial clients either, right? It has to, the math has to close on what the payback period is on the industrial side. And it has to close, but the budgetary concerns on the DoD side, right? real defense budgets are in, in You know, real dollars down.

And so, we, we look at that being a big advantage of like we're gonna tailor our hardware stack and all of the underlying safety goes with it, specifically to the software we're building. And because of that it looks more like a Mac solution than a Alienware Linux machine. We're like, we have a bunch of Alienware Linux machines around here.

They're great. And then you can do a lot of stuff on them, but 99 percent of the time, those things are running at like 5 percent load. Um, andyou're, again, your customer doesn't care what that hardware looks like as long as it works. And so you have to optimize for reliability and safety, manufacturability.

Can you produce it in time? Does it take two years to get the parts you need for something, or can you do it in 16 weeks? I think that it's one of those spaces. And I think the OS manufacturers of the world that actually have market share, i. e. Mac and, or Apple on the Mac platform and windows would argue that tightly coupled software and hardware actually lead to a much better user experience.

because you know exactly what you're building to versus. Okay. What is this? Oh, yeah. I'm going to put on the vehicle. Or what is this? What is this? You know, I'm going to hand you a stack from the DOD and say, Hey, go run this here.we've seen customers who are like, yeah, we've got plenty of compute on board and it's like an I seven laptop effectively.

Like let's say you have a government issued laptop and you're great. This is like probably not going to work. Like you're going to need some GPUs to do this. so we think those things have to be tightly coupled. And we think that's an advantage of being a relatively small company is that the hardware and the software engineers are, In the same space and they can develop requirements together and then go over to production and say, this is what we want to build.

Is this actually something we can design for and manufacture and, get through our CMs in a reasonable amount of time. And that's, we think it's really important to be able to deliver autonomy at scale, and at a readiness level that you can actually go use it.

[00:10:06] Ryan Connell: No, got it. And I like what you just hit on there. What's that secret sauce look like in terms of. So software engineer, hardware engineer, production, like, is there a code or a hack for having, a certain group together that kind of worked through that? and if so, is it just those three people or is there more than that?

[00:10:25] Scott Sanders: There tends to be more than that, but in theory, like these cross functional teams are just that you need a representative from,all parts of the engineering teams that are touching this and the production and operation teams that are building it and the, business development and growth and account management people who are, putting their names on this product and talking to their customers who,we have a large number of vets at this company.

We often are friends with the people we're selling it to, or it's going to our friends, you know, people who are our friends and that shouldn't matter, whether they're a friend or not, whether your product works, but like, there's a, it hits home way more when someone that you know has to rely on your product, it's not just some

faceless PEO or Pm. It's like there's an actual, captain that you serve with in the Marine Corps is taking this into combat someday, or it's one of your like Lance corporals who's somehow got promoted or and or became an officer. and are now taking your tech into the field. It has to work.

And so those cross functional teams are aligned around, the growth side and the engineering side, and they have to work together to deliver a product. You can't do it with just one.

[00:11:37] Ryan Connell: And you brought up an interesting point on price in terms of just like pricing at a fair price that DOD finds value. have you in your experience noticed any, mindset differences in your commercial world space versus DOD? And I don't want to give away that the answer that I think it is, but I would imagine that maybe in the commercial space, it's much more like ROI based.

Like we think we're going to add X amount of value by pursuing that investment. I would have guessed that maybe in the DOD space, it's, we want to pay for three bodies times this rate. And I don't know if that's something that you're, dealing with,

[00:12:08] Scott Sanders: be cool. in the industrial space, you're spot on. It's well, when I get to this ratio of autonomy, and this is what the head count costs, to do that work, especially for enterprises that are running three shifts per day, the payback periods like relatively quick. And we think that's the valuable part about becoming a platform in the industrial space is that it's not just the value that we take, as a company, cause You know, say the quiet thing out loud, especially in the defense space, like you, you have to make a profit.

if you don't, you're not going to last very long. or at least have a glide path to get there. andyou, but you can't take all of it, right? If you have a deal where you'll say you're saving a hundred thousand dollars per year per vehicle. if you, the autonomy provider is taking 90 percent of that's not particularly interesting to your industrial partners, right?

That's gotta be somewhat fair. And that's hard because the cost of labor. Is different between a one shift operation and a three shift operation. And is that operation like in the middle of nowhere and they've got to pay a big premium to get people there? Or is this in a place where there is actually a lot of labor available?

So finding the right price, was something that we put a lot of effort into and making sure that it's fair. and that, you know, both, we can both run a profitable company off of it, hopefully someday and our partners are getting the value out of it, that it's actually interesting to them. calculations tend to be a little bit different where it's not just the labor costs on the industrial side.

It's also, Hey, you're backing up these giant 40 foot, trailers that have, one might have a hundred thousand pounds in it. One might be empty, and those things back up differently. And so, uh, A human driver can often struggle to do that, and, we don't give them a lot of tolerance space inside of the yards and the ports, they're like very narrow, like avenues these vehicles have to operate in, and they're not particularly strong trailers, so if you want to go on YouTube and just like, Backup trailer fail search.

You're

[00:13:51] Ryan Connell: Yeah.

getting ripped apart by ballards. My evening is planned.

[00:13:55] Scott Sanders: it's a good one. And if you just you get more bored, just go like RV backup fail. Um, and so you have the warranty costs, you have the hiring costs, it's supposed to be an area with high turnover. if you're reliant, if you're running 24 hour, a daily shifts and people just don't show up to work, that's Margin,for those businesses.

So it's predictability, it's consistency, on the DOD side. haven't seen anyone do like math yet on, on why it should cost a certain price, but I think it's also way harder. if you want to go on a much more boring, Google search operation, go look and see how many times Rand is trying to figure out what a soldier actually costs based on rank.

It is An unknown. Um, well, what training did they go through? How long have they been in? Did they get injured when they were in? Like it's anywhere between 400 K and 5 million. there's some number in there. nevermind the fact that it's a human life. And there's the easy math of what the SGLI policy costs and what are the entry level training costs.

But when we think more critically about the wars that we will potentially fight in the future, what it takes to replace. That person isn't just the dollar value like we don't have the stomach to go to combat and lose people and we shouldn't It's like pretty terrible But also if you're gonna be fighting and say in no paycom, you're not getting a replacement human out there Like until someone amends a underwater aircraft carrier Nuclear submarine, which we built in 2046, you're not resupplying those people, not with any kind of mass.

And so it's super critical that the humans that we do put forward are as lethal as possible. so I think, the DoD is trying to figure out what they think the price is. We offer the same price to the defense community as we do to the commercial community, because we sell a commercial product, so it makes it a little bit easier for us.

And, our goal, crazy enough, is to get the price to go down, which is, not a thing you hear defense contractors talk about. because our logic is, Commerciality cuts both ways. If I'm selling it for one price on the commercial side and the same price on the defense side, in order for me to address more commercial markets, the price has to come down and it's got to come down a lot.

and so I'm incentivized to put all of my investment into how do I make that hardware stack cheaper? How do I address a bigger market? So I can amortize the cost of developing software over a longer period of time. And as I want to increase my commercial market penetration and. Bring that cost down. I have to pass those savings along to the D.

O. D. and we think that's a great way to do business. It's a fair way to do business. and at the end of the day, we don't care whether it's a purple pterodactyl or autonomy stack as long as there's not a human in it. And sadly, there's no purple, purple pterodactyls left around. but there's really no reason to be losing lives when we have the capability not to.

And autonomy is not going to solve every problem today. And I would guess in the future it doesn't solve every problem either. You're going to have humans on the loop. The question is, is it one to one? Is it one to four? Is it one to a hundred? The only way to figure that out is to get systems out today and have people actually use them for a sustained period of time versus just limited scale soldier touch points.

so we're lucky to be a part of programs that have Systems that are going out and being fielded in mass. And we think we're going to learn a lot from that random Lance Corporal. He was like, Oh, you just gave me a robotic missile truck. sweet. I'm going to show you exactly what your engineers didn't find in a regression testing.

No, that's, that's really cool. Yeah. And so what made me think from my perspective, a lot of what I do day to day is more related to, acquisition function use cases, right? So not autonomy in terms of missiles, but think like AI market research, like that kind of thing. Right.

[00:17:24] Ryan Connell: And that's what you just identified has been the exact same challenge For me is like, does the autonomy or the AI market research, it doesn't replace the people. it just gonna augment their team. So I can't say, Oh my gosh, I no longer need three people. I have this other thing. It's I no longer need.

I don't know, 10 percent of their time,So then coming to that valuation of what it's actually worth. That's I've struggled with that, to be honest, from a buying side. Yeah. Yeah.

[00:17:57] Scott Sanders: costs in the DoD, like, please let us know, that'd be great for doing Excel math. But I think it's a little more cut and dry on the autonomy side. Like you can look at a convoy of like heavy trucks in the army or the Marine Corps and say, okay, I used to do this convoy with 40 people.

Now I can do it safely with two and that's, 38 less people that I have to

eat 

[00:18:18] Ryan Connell: very clear. Yeah.

[00:18:20] Scott Sanders: right? And there's stack, there are additional costs that stack with humans. you've got to, you got to feed them. They need water. They need a place to sleep. They need security. And you create the tooth to tail that we have today in the U S D O D where it's like pretty out of whack.

most of the force isn't combat Marines or, Infantry soldiers, it's support. And if we can take that force structure and give it back to the operating forces. not that any, not that, all the jobs in the military are very important, but like the end of the day, someone is clearing a trench, unfortunately.

[00:18:53] Ryan Connell: That's just how warfare works. And so the more we can enable those functions, we think that's really important. And also no one likes riding in trucks, at least I never did. sure.

I can't see what's going on,With, I like that example. I think that's, it's cut and dry kind of black and white, anecdote. and so we talked price a little bit in valuation and obviously that kind of dovetails into like risk. and I'm curious, what do we do? We meaning you and D. O. D. in terms of, understanding the risk associated with.

So I think you said 40 people and then going down to 2 with some autonomous systems and reducing 38. I can kind of in my head imagine that there's, there's risk of people that, people are, Do things that probably shouldn't do and all of that. And that comes with its own,human nature element of that.

And then there's risk of the autonomous system potentially having a challenge. are we all in agreement that the assumption is the risk is also less when we go, with autonomy, or are we actually increasing risk or what does that look like?

[00:19:50] Scott Sanders: do you ever read that Nate Fick book back in the day called One Bullet Away?

[00:19:54] Ryan Connell: I didn't, but I'm writing it down.

[00:19:56] Scott Sanders: Alright, that's a good one. we used to joke, so that came out when I was a young platoon commander. we all used to joke internally that the reason it's called One Bullet Away is that Nate was like, as a rifle battalion, or rifle company XO, he was one bullet away from being in charge.

We used to make the joke that you're always one Lance Corporal away from getting fired. Because the Marines, they do the darndest things. Um, you're like, oh, you decided to buy a Gixxer 1000 and drive up to five going 120 on the weekend just for fun. Interesting. Goodbye week. Like I'm going to be in this

[00:20:28] Ryan Connell: That's what I mean. That's what exactly I'm talking about. Yeah. Yeah.

[00:20:31] Scott Sanders: Okay, that's, that was neat. your life's over. Congratulations. Mine is too by proxy, because I, I didn't brief you that you shouldn't do that on the weekend. and that, that should resonate with most people who've led, our nation to war fighters in the combat.

It's just man, you guys are awesome when there's fighting, but the no fighting is terrifying. Um, and so I, I think you're trading risk. And I think that's a hard thing to figure out because there isn't. A UL 4,600 an ISO 2 6 2 6 2, or even 8 82 Echo, that's gonna help you figure that out. andI look at like, where are the highest risk scenarios?

It's anytime people leave the wire and go into combat or you're doing a logistics run, and it's pretty easy to go look at the non-combat death in the US military. It is all vehicles so it's like you have. These are already high risk operations.

Anything we can do the lower the risk. And I think the thing that us military is incredibly good at, and probably doesn't get enough credit for his operational safety, like we know how to employ weapons systems. We just keep people away from them. It's you want to drop a bomb? Cool. You just need to be a thousand meters away.

It's like pretty far. And that's what you do in training. And if we get to the, go to the combats, we're like, All right. With that bomb, we'll give you 250 meters. that's way closer. Are we sure that's all right? yeah, math on it seems right. hopefully the pilots paying attention. Hopefully you gave them their grid, the enemy grid, not your own.

like there's some checks and balances on this. So I think autonomy is no different. It's like you can deploy it stateside with enhanced levels of operational controls, and then give that captain, major, staff sergeant, gunnery sergeant, give them that ability to say, okay, I'm willing to take more operational risk knowing that Anytime I send people out and out into combat, I'm exposing myself to a great deal of risk.

And if I can just reduce the amount of surface area of humans, that's going to be better. And I think it's also more pronounced than OEF OIF with great power competition and that we're going to be fighting an enemy. that is like pretty capable. and there, we used to worry about mortars being dropped on bases in Afghanistan.

And now if you're still deployed to Iraq, you're worried about small UAS, medium sized UAS, Shade 129s, things like that, but nothing that compares to, I just shot an anti ship ballistic missile at a thing that is basically a copy of a U. S. DDG. And it's going to be shooting back within a minute because finding something that's doing Mach 4 on the battlefield And it's like the size of a telephone pole might be one of the easiest targeting problems in the history of targeting.

There's not a whole lot of things that look like that in the air. And if there is, you have other problems. Um, get a telephone pole drained down on you? Like, where are you? so by pulling people out of those platforms, you're just, you're reducing, you're not reducing the material risk. your robots might get shot.

Your robot might only cost a million bucks. And it holds six telephone poles of,Gimler's rockets. And once they're all shot, if they get shot, it's not that big of a deal. it's a big deal. I think most people have flippled worse, if you write off equipment. but you're not having to replace that human and that human that we just talked about is, you know, somewhere in the Indo Pacific, you're not getting a replacement human.

At least not anytime soon.

[00:23:51] Ryan Connell: Got it. Had a note to just talk and we talked pricing and risk and all these things, but just kind of acquisition in general. anything that you wanted to,is there anything that we do differently with autonomous, acquisition or is there anything we do differently in DoD versus your industrial customers, that you wanted to highlight?

[00:24:07] Scott Sanders: Yeah. You, you asked a question earlier about highlighting the difference between the two, especially on pricing. I forgot to mention my favorite part, which is you talk to the DOD, you say, this is the price. And they're like, your commercial clients, they can afford that super expensive to be clear.

It's like we're talking, I pay more for Salesforce in the commercial space. It's not that expensive. And then, you know, they're like, Oh, but the, the fortune 500 companies in the world, like they can afford that.

You're like, Oh, okay. You go over to the fortune 500 companies and they're like, some of, some of them are like, Hey, it's a fair price. Some are like, that seems high. And you're like, yeah, it's like, here's the spreadsheet math. We think this is very fair. And most of them come around, but they always start with, the DOD can pay that.

And like, you guys are just the same, like you, you have to come up with a financial model of where, like, how does this make sense? And I think most companies, given if there was enough pushback. Would reevaluate the price, but it's also a function of what did it cost to develop this? And how can I amortize recouping those costs over time?

Because again, it's as fun as it would be to run like a DoD charity. Like I'm gonna say like 5 billion laying around and I could just go do whatever science projects I wanted to and didn't care about making money. That'd be a great world. That's not a world they live in. you have to have a foundational business model for this to close or these companies are not going to be around.

and I think when it comes to the acquisition side, I don't think there's, we don't really do anything different. and I think that a lot of earliest days of defense tech companies get on podcasts or panels and just kind of like tear their customer apart for being bad at contracting. And to be fair, there's plenty to tear apart, but you need to learn to work within the system and develop a product and a way to sell it that's going to work within the DOD system. Because you're not going to change it. I would also argue that like the rules are fine. Okay. The FAR is designed to be fair both ways, and it is, and OTs are great, but I do a FAR 12 contract every day of the week over an OT because it has a set series of rules that you can follow, and they make a lot of sense, I think the hard thing in this space is a similar problem that Major League Baseball had, which is the rules were fine, but the people calling balls and strikes can be hit or miss one contract to the other, and so, you know, what did Major League Baseball do?

They digitized a lot of it. And you could probably digitize all of it. I think that there has to be a better way to fix acquisitions, the wrong thing, but they need better tooling to move quickly and also feel comfortable with what they're,what they're signing their warrant against. because we do put a lot of trust and confidence in PCOs, KOs, and AOs to act on behalf of us as taxpayers.

In the best interest of the U. S. government. It's a lot of responsibility to hand a GS 9, right? hey, here's a half billion dollar program line. Go execute all this. Oh, if you screw up, by the way, we're gonna ruin your life. so I think there could be a bigger focus on using things like AI to enhance their quality of life and ability to Really understand what they're signing up for into your point on that's things like market research.

That's things like, you know Why do we have 19 different platforms that you need to be proficient in to be a KO? every single servants does this different and every single contracting office does this different and so having the easier way to have commonality in a common sort of workplace, I think would actually do a lot in a good way.

It's definitely not an area I've seen anyone, go pitch of like, Hey, I'm gonna go do this startup. And I'm just going to make, I'm just going to focus on doing like an ERP for defense and make it easier for contracting officers to get on contract. Instead, we're like, Oh, let's go lobby Congress and change the rules.

Like how the rules seem fine. Like they just need to be able to use them. And they're, we didn't make them simple. there's a lot of them. I

[00:27:49] Ryan Connell: Yeah. Yeah. What do you got? 1300 page far and D far each. Something like that. 

[00:27:53] Scott Sanders: it's more than that. 

[00:27:56] Ryan Connell: I just, I'm reading a book and they said 1300 for defars. I don't know what far is. no, very cool. maybe one last question here. I'm curious, being a small company and being able to find your way and navigate your way.

do you have any experience or lessons learned on navigating defense acquisition as like a small business trying to navigate like the barriers to entry and all of that. so I guess maybe experience and, or advice for anyone that's in a similar situation as you.

[00:28:22] Scott Sanders: I think I'll start with advice, which is, this is pretty straightforward, but the DoD is not a single market, right? It's a lot of micro markets that have adjacent requirements to each other. and so there isn't this like a DoD ground autonomy market, just like there's not like a DoD drone market.

There is different elements in the army that have different requirements for drones and,they are different than the air force and they're different than Marine Corps and yep. Yeah. SOCOM is probably different too. I think it's understanding that there is no playbook. There is no set series of you just do these steps like you would an enterprise SAS and eventually you will get sales.

I think you have to have a really unique understanding of what is the requirement, like what, or is there not one, or does it need to change? Or was it written too narrowly? which I think is actually one of the things that, that DoD in general could do better. It's just treat it more like air power.

Ask for an effect. Let the nation's best engineers try to go give you the best solution for that and then vigorously compete that in the field.as for,experience, experiences may vary. This is a very complicated space and there's a lot of ups and downs as you work these problem sets.

and I think that a lot of people want to come in and They want to show that they can sell a capable product to the DoD, not realizing that,DoD sales cycles are outside the length of your venture capital cycles. So you need to figure out how to do that. To layer in kind of your small entry level, foothold sales with your, I need to develop a program, a record for this capability.

And one doesn't exist or one does exist, but it's going to be competed in 2027. Um, and I think it's just, it's very different from the commercial side, in that regard,commercial sales can take a while too. You're dealing with big companies that are not as big as a DOD, but they're like a 10th the size, like it's still pretty big.

Um, so at the end of the day, if you work in this space, there's one big, one really big upside, which is you're building something to help, preserve Western values and Western democracy, and making better emojis at Facebook's probably not the place to do that.

although the new sticker thing that Apple came out with is pretty hilarious. 

[00:30:32] Ryan Connell: I haven't dabbled. I got to get on it.

[00:30:34] Scott Sanders: it's really great for

[00:30:35] Ryan Connell: Okay. Okay.

[00:30:36] Scott Sanders: it's about,

[00:30:38] Ryan Connell: Awesome. Scott, Hey,any last thoughts before we wrap here?

[00:30:42] Scott Sanders: No, thanks for having me on and, getting a chat.

[00:30:45] Ryan Connell: Yeah, Scott. Appreciate you being on. Thanks so much. Talk soon.