#234 Tyler Xuan Saltsman: How AI is Shaping the Future of Combat & Warfare

In this episode of the Eye on AI podcast, Tyler Xuan Saltsman, CEO of Edgerunner, joins Craig Smith to explore how AI is reshaping military strategy, logistics, and defense technology—pushing the boundaries of what's possible in modern warfare. Tyler shares the vision behind Edgerunner, a company at the cutting edge of generative AI for military applications. From logistics and mission planning to autonomous drones and battlefield intelligence, Edgerunner is building domain-specific AI that enhances decision-making, ensuring national security while keeping humans in control. We dive into how AI-powered military agents work, including the LoRA (Low-Rank Adaptation) model, which fine-tunes AI to think and act like military specialists—whether in logistics, aircraft maintenance, or real-time combat scenarios. Tyler explains how retrieval-augmented generation (RAG) and small language models allow warfighters to access mission-critical intelligence without relying on the internet, bringing real-time AI support directly to the battlefield. Tyler also discusses the future of drone warfare—how AI-driven, vision-enabled drones can neutralize threats autonomously, reducing reliance on human pilots while increasing battlefield efficiency. With autonomous swarms, AI-powered kamikaze drones, and real-time situational awareness, the landscape of modern warfare is evolving fast. Beyond combat, we explore AI's role in security, including advanced weapons detection systems that can safeguard military bases, schools, and public spaces. Tyler highlights the urgent need for transparency in AI, contrasting Edgerunner's open and auditable AI models with the black-box approaches of major tech companies. Discover how AI is transforming military operations, from logistics to combat strategy, and what this means for the future of defense technology. Don't forget to like, subscribe, and hit the notification bell for more deep dives into AI, defense, and cutting-edge technology! Stay Updated: Craig Smith Twitter: https://twitter.com/craigss Eye on A. I. Twitter: https://twitter.com/EyeOn_AI 00:00) Introduction – AI for the Warfighter (01:34) How AI is Transforming Military Logistics( 04:44) Running AI on the Edge – No Internet Required (06:49) AI-Powered Mission Planning & Risk Mitigation (14:32) The Future of AI in Drone Warfare (22:17) AI's Role in Strategic Defense & Economic Warfare (26:34) The U. S.-China AI Race – Are We Falling Behind? (35:17) The Future of AI in Warfare

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Our logistics agent is used with a meta model fine tuned on a bunch of military logistics doctrine. And then now that model now thinks like a military logistician. And what we do is we build these adapters, they're called LoRa, which stands for a low rank adaptation of a large language model. And then now what you do is that adapter freezes 99% of the parameters of the model. And then now the model only speaks to the information pertinent in that adapter. So rather than a generalized chat model, which is what we see today that are pretty much unusable. It's now specific to a logistician. Tyler Saltsman here, I'm founder and CEO of Edge Runner. We're building generative AI for the warfighter. And what that means is we're building domain specific intelligence.

• 0:43Speaker 00

  That's occupation specific for each, each, um, each warfighter's role. Whether it's logistics, condition based maintenance, uh, whether it's, um, you're a fighter pilot, you operate a battleship or a submarine. We're building AI that'll be, that'll, that'll augment all of our warfighters to ensure national security. And what also makes us different is our AI is open and auditable, meaning you can understand the training data. And the biases that went into the AI, unlike our peers today.

• 1:17Craig S. Smith?

  That's interesting. And, and what are the primary models and use cases?

• 1:24Craig S. Smith?

  I mean, when you say logistics, I mean specifically or, or, uh, you know, someone, uh, managing a, a weapons platform.

• 1:34Speaker 00

  Yeah. So right now we're in a crawl, walk, run. The, the crawl phase where you, we're actually excited to partner up with Meta. So we have clearance to use, uh, Metas Lama models for DoD applications. So our logistics agent, um, is used with a meta model fine tuned on a bunch of military logistics doctrine. And then now that, that new, that model now thinks like a military logistician. And what we do is we build these adapters, they're called LoRa, which stands for a low rank adaptation of a large language model. And then now what you do is that adapter freezes 99% of the parameters of the model. And then now the model only speaks to the information pertinent in that, in that adapter.

• 2:16Speaker 00

  So rather than a generalized chat model, which is what we see today that are pretty much unusable. It's now specific to a logistician.

• 2:25Craig S. Smith?

  And this is a fine-tuned model. You're not using a RAG system.

• 2:32Speaker 00

  We also use RAG. So I like to think of what we're doing is we're building,

• 2:40Speaker 00

  we're building a combination of a RAG pipeline, of LoRa adapters, and with small language models that all work together synergistically. So for example, you would use RAG on things that are static and that don't change. And then you would use a small language model with a LoRa on something that's dynamic and that you need to be creative. So here's a good example. Let's say you're a fighter pilot and let's say I need to know everything about my fighter jet. How do I maintain it? How does it perform under certain circumstances?

• 3:11Speaker 00

  An interactive manual on steroids, if you will, that I can talk to. That would be your RAG pipeline because RAG, Retrieval Augmented Generation, is just retrieving all the relevant information of that fighter jet. But now, let's say I have a bespoke mission where I'm doing something that's not common. Let's say I have to do an air-to-ground mission and I'm flying low to the ground with an F-22 or an F-16. That would then be a small language model with a LoRa adapter attached to it. So I'm using a combination of RAG when I need to interact with my fighter jet. And then I have my small language model with my LoRa for the bespoke mission that's specialized.

• 3:48Craig S. Smith?

  Yeah, talk about how that small language model with LoRa would work. And is that on the edge? I mean, is that in the aircraft?

• 3:59Speaker 00

  That's right. So for example, what's great about LoRa is its access is fine-tuning on the edge. Because we're freezing basically 99% of the entire model and we're only speaking to what's in that adapter. So that pilot would then have an adapter that's specific to the terrain, the demographics, the mission here she is executing, the target that they might need to go take down, or maybe it's a reconnaissance mission. Basically, all the situational awareness you'll need will be in that LoRa. And then the small language model from Meta, for example, would then be augmented by it. So we would build that LoRa, we would run that Meta model, we'd also provide the RAG pipeline, and that would all work right on a device on the edge that doesn't need the internet.

• 4:44Craig S. Smith?

  And what, how does the pilot interact with the model then?

• 4:49Speaker 00

  Right now, right now, the pilot would, they would, they would type in the device, but we're in the process of making this voice enabled, kinda like Jarvis or Iron Man.

• 4:59Craig S. Smith?

  Mm hmm. And, and what, what's an example of what a pilot would be asking the model for?

• 5:09Speaker 00

  Yeah, let's say, um,

• 5:13Speaker 00

  let's say the pilots noticing something unusual sound within the aircraft. Hey, what is this? Is this a problem? Do I need to divert and ground the aircraft? That could be something that comes up, or let's say, let's say there's a, there's an enemy on, you know, on the horizon. You know, based on my payload, what I have, do I engage the enemy? What do I do? What kind of evasive maneuvers can I make safely? Based on, you know, how the, how that fighter gets configured, something like that. Um, and again, I don't, I don't know much about flying, but I could get into more like logistics, which is what I know. I used to be a logistician for the army and we also prepared for you a logistician, a demo of our technology.

• 5:57Craig S. Smith?

  Sure. Yeah. Well, let's talk about that because logistics is a,

• 6:03Craig S. Smith?

  is a big piece of the military.

• 6:05Speaker 00

  That's right. We see in, in Russian-Ukraine conflict right now. Um, Ukraine sometimes gets the upper hand because of better logistics. And so of course, if you run out of fuel or life support or ammo, you name it, your fighting force is down. And so we have to make sure that we're optimizing for that. So for example, with logistics, it can help me like which ammo goes with which ammo on which truck? What happens if this truck goes down? What alternate routes do I have? Um, if I get ambushed, what are my, what are my friendly units around the area? Um, you know, if, if I'm hearing a weird noise in this truck, what, how do I troubleshoot it? What's wrong with it? So AI can detect all of this and help me on the fly right away.

• 6:47Speaker 00

  Um, but I'll pause there. Colton can bring up the demo. Let's, let's walk through that and show you.

• 6:52Craig S. Smith?

  Sure.

• 6:52Speaker 00

  The actual working prototype. And again, this works locally on device without ever needing the internet. So we're going to show that we're going to turn on or turn off wifi.

• 7:04Speaker 00

  Of course, this is just a demo, but here you can see a very comprehensive prompt. For example, if you ask ChatGPT this or any kind of model, it wouldn't know this. Because again, these models are too generalized. Now, what we're able to do is drill into this exact mission ask or the prompt. And how do we properly plan a logistics mission of this magnitude? And how do we mitigate risk? What's the equipment we're going to need? You know, what is the planning phase look like versus the execution phase? You name it. So we can get very granular. And this helps a logistician officer like me better plan and, you know, and do my job.

• 7:45Craig S. Smith?

  And what are the PDFs listed below?

• 7:52Speaker 00

  So what we do is we actually cite all the sources that we trained on. So now, if I say, okay, this is an interesting line item and the recommendation the AI gave me. Where did that come from? Now you can drill into that, that the doctrine of where it came from.

• 8:08Craig S. Smith?

  Yeah. Yeah, this is remarkable. And for audio listeners, we're looking at at Edgerunner, the logistics agent.

• 8:25Craig S. Smith?

  Built on is that, how do you say that? It's, it's, it's an instance of Edgerunner or a use case of Edgerunner.

• 8:33Speaker 00

  That's right. And what this is, so we're actually showing you all the different sources we trained on. And we trained this on thousands of different PDFs. Related to the Army and logistics. So it's very specific.

• 8:48Craig S. Smith?

  And it's giving a breakdown of

• 8:55Craig S. Smith?

  day by day, you know, morning, midday, late afternoon, exactly what has to happen.

• 9:03Speaker 00

  Exactly like this, this what it just generated for me in literally 30 seconds can take, could take me hours or even, or even a couple of days. And so now this can augment, and this is based on like the Army doctrine way. And then, of course, you know, things in the battlefield aren't always going to be by the book or by, by doctrine. But at least now it's giving me a very good baseline of what the mission should look like.

• 9:28Craig S. Smith?

  And and before something like this or or currently it's it's all done by hand or are there systems that that are doing this without generative AI?

• 9:42Speaker 00

  So this is completely not not not not done by hand at all. But what you can do is now that we have like this baseline, we could then go whiteboard and have an art of the possible and maybe we want to take out day two so then we can just quickly modify it via hand jamming.

• 10:01Craig S. Smith?

  Right. And what's the the uptake at this point? I mean, where are you guys in in your? Give me a little background on on the company when you formed and

• 10:16Speaker 00

  Yeah.

• 10:17Craig S. Smith?

  where you are, yeah?

• 10:18Speaker 00

  Yeah, so the genesis of us, so we're all pretty much extability AI. Instability was it was known for stable diffusion, which is the most popular, one of the most popular media models in the world for for text to image modality. And from, you know, in in in working locks up with all these scientists, you know, we started to because I was I was had a supercomputer when I was at stability. And working with these scientists, we realized we had a core competency of being able to train and inference models across any different types of of chip architecture and hardware, but also to make these models much smaller. And then they and when they're smaller, they can live right on the device and not need the Internet. I realized we had this core competency, you know, it's like, let's go build this for

• 11:02Speaker 00

  the military, you know, and given my background as as an ex-army officer and also meeting meeting these scientists, we just it just made a lot of sense for us to work together. And so it was challenging to at stability because we weren't really creating products that added value to customers. It was more just a cool thing, but there was no real value creation. I think that's the problem with AI today. It's not actually solving anyone's problem and it's not really doing anything useful other than making cool things. So now we were solving for that by actually building products that our customers, the DOD, that they could use.

• 11:41Craig S. Smith?

  Right.

• 11:41Speaker 00

  Um, so we're only about eight months old. We raised five and a half million. Uh, Madrona is the biggest VC on our cap table. Uh, they're, they're an amazing partner. Matt McElwain has been awesome supporter of us as well as John Turo. Um, and then of course we have some angels. Um, so yeah, uh, we have a full team of 18. And, um, we're, um, starting to ramp up to scale early next year.

• 12:06Craig S. Smith?

  Uh, you know, I was, I worked on the National Security Commission on AI. And for a couple of years, the, the two years that it ran, uh, and, and there was a lot of talk about integrating uh, AI into, I mean, that was the focus of the commission, really how to accelerate that integration. Uh, is there anything, uh, I mean, how does this work? So you guys, um, you know, you, you have a company and a product and relationships with the DOD.

• 12:44Craig S. Smith?

  Uh, it's so complex, the, uh, the acquisition, uh, process. Uh, how, how, how do you manage that? How do you navigate that?

• 13:00Speaker 00

  You know, uh, we're, we're going to be going through the channel. Uh, right now we're on Carisoft. They're one of the biggest, um, channel distribution partners of the DOD. So we're a proved vendor on Carisoft. We're on NASA soup as well as, uh, ITES SW2 contract vehicles that the army uses. But the DOD uses NASA soup. So we're, we're good to go. We're on that vehicle. So when, when you need us, you can procure us directly through them.

• 13:27Speaker 00

  And then of course this will expand. Go ahead.

• 13:30Craig S. Smith?

  And, and then it's up to the individual unit commander or who makes that decision about whether to? Uh, to onboard Edge Runner.

• 13:39Speaker 00

  Yeah, that so that'll come way up at like the at the J level, even at the Pentagon. Typically, it'll it'll come down from the top if there's some sort of AI mandate or policy. Then there'll be what we're going after is called broad area announcements. There'll be broad area announcements of AI to do XYZ. And then we'll then we'll then bid on it. But what helps us is we're actually for us, we qualify as a small business and we're veteran owned. And I have a PTSD disability rating. So we um, we're a strategic partner for these bigger partners that are required to work with smaller guys like us. And of course, we have core competencies of building AI that's personalized and completely air gapped. Um, which actually segues into our computer vision technology.

• 14:24Speaker 00

  So not only are we just are we building agents and assistants on device? But let's show you our computer vision technology.

• 14:31Craig S. Smith?

  Sure.

• 14:32Speaker 00

  And what we're solving for is right now in Ukraine. Um, the life expectancy when you're when you're spotted by a Russian drone is only seven minutes.

• 14:41Craig S. Smith?

  Wow.

• 14:42Speaker 00

  Until you find cover and concealment. And so how do we neutralize these Russian drones? Well, um, we're we're building drones that can see now rather than requiring a pilot. Because when you require pilots, easy to jam these drones and disconnect the pilot from the drone, rendering the drone useless. But now we can put vision on swarms of consumer grade drones, which are very cheap. That then carry a payload via a kamikaze suicide drone style. So now we'll show you um, what that computer vision technology looks like and it works today.

• 15:22Speaker 00

  And that's the problem that we're solving for is it's easy to jam these rounds. But now when they can see like what you're seeing now.

• 15:29Craig S. Smith?

  Yeah.

• 15:30Speaker 00

  These are autonomous and they can fly right into that tank and with a payload and and neutralize it.

• 15:36Craig S. Smith?

  Uh, and how how do you ensure that it's not going to fly into a Ukrainian?

• 15:46Craig S. Smith?

  You know, piece of Ukrainian armor.

• 15:49Speaker 00

  No, that's a great point. It's, it's, and it's what we're doing is we're building the neural network. Just like a human, just if a human can discern the difference between a Ukrainian tank and a Russian tank, AI can as well.

• 16:01Craig S. Smith?

  Yeah.

• 16:01Speaker 00

  You do that by training the model on, you know, hundreds of thousands of images.

• 16:05Craig S. Smith?

  Yeah.

• 16:06Speaker 00

  Now, to your point, if they look identical, then you're right, we can't. So, so if Russia now rolled out tanks that are identical to Ukrainian tanks, that would be problematic. We'd have to find another solution. But right now, it's very easy to discern what a Russian tank looks like versus an ally tank.

• 16:22Craig S. Smith?

  Yeah. Okay. And so again, for audio listeners, uh, that, that don't want to go over to YouTube, uh, to see the video, uh, we're looking at, uh, footage from a drone or from a series of drones, uh, hitting Russian armor. Uh, is this training data or is, is, uh, you don't have these drones fielded these, uh, yet, do you?

• 16:49Speaker 00

  Correct. So the drones aren't deployed in real life combat zones yet. Right now, we're testing. And this is an overlay of our technology from drones that can't see. Now we're showcasing that they could see. So this is more of the art of the possible. But the technology does work and we're in the process of, um, of bringing this to life with, with our partners. So we're working with Service Capital as well as, uh, the, and the drone company is called Rapid Flight. And their, their core common sense is they build 3D. Uh, it, it's easy to, they, they print drones via a 3D printer. And then these drones are, are one-way drones that can, that are cheap to make. They can make them quickly. And then, of course, they can carry a payload.

• 17:31Craig S. Smith?

  And where are those drones made?

• 17:33Speaker 00

  They're made, uh, they're made here with, with our, our partner called Rapid Flight.

• 17:38Craig S. Smith?

  I see. Are, are you familiar with Eric, what Eric Schmidt's doing with White Stork? I think it's called White Stork.

• 17:45Speaker 00

  I know Eric Schmidt's got some really cool projects going on that I'd love to partner with him on. Um, so I'm not familiar with that exact project, but I know that Eric has a, a keen interest in, in, in what we're working on as well. So if you happen to know him, please give him a shout.

• 18:01Craig S. Smith?

  Yeah. Well, I'm hoping to have him on the podcast soon, so I'll mention it to him. But he's very interested in, in drone technology, just, uh, in how it's, uh, changing.

• 18:12Craig S. Smith?

  The nature of warfare. I mean, it really has. I mean, what do you think it if all the analysis I've read is that, you know, armor is is pretty much obsolete at this point?

• 18:27Speaker 00

  That's right. I mean, as drones become ubiquitous, it's definitely going to change how we fight wars. Now we're going to have drone on drone combat. Um, also, what will be interesting too is AI, large language models still aren't intelligent yet. And I think largely it's because they don't understand the physical world. Well, how do we understand the physical world? We've put VLMs on drones and drones now scan the battlefield and are describing everything that they're seeing. Now we can take that new synthetic data and reinforce the neural network of the large language model, which then makes the on-device agents better. And now we get another step, another great leap forward of AI. Now, I don't think this will create AGI, but I think it'll make generalized AI much better.

• 19:12Craig S. Smith?

  Yeah, the and you'll forgive me, Tyler. I don't remember whether we spoke about this. Were we talking about coordinating swarms last time we spoke?

• 19:28Speaker 00

  A little bit. And I think the

• 19:33Speaker 00

  the course on the conversation was around that humans will always be in the loop.

• 19:36Craig S. Smith?

  Right.

• 19:37Speaker 00

  We'll never allow, we'll never want AI to be the shock collar, if you will. I think AI should be great at the expert assistant that's advising you objectively and fairly, right? But never actually making the decision. Kind of like Jarvis in Iron Man. If you watch the Marvel movies, Tony Stark's always ignoring Jarvis, even though Jarvis is probably right. But he still has the ability to ignore Jarvis. Because I think there's still the human gut element that we need to trust. But of course, AI is going to make us better at what we do.

• 20:12Craig S. Smith?

  Yeah, although a human in a loop, how do you explain a human in the loop with with a drone that that is, you know, is vision enabled and?

• 20:26Craig S. Smith?

  A Kamikaze drone that's, you know, navigating its way on its own.

• 20:32Speaker 00

  I think what we do is, so we can train drones to see hand and arm signals. So I can wave to my drone to come here, or stop, or go, things like that. So now, we call this being context aware. So normally, a drone can just see a human, and it's not recognizing the symbols I'm making with my, my, my hands. But now that the drone's context aware, I can, I can effectively control it. You know, kind of like soldiers, I can't talk to my soldier, but if we need to go attack, I can use certain hand and arm signals and we can go engage without talking.

• 21:06Craig S. Smith?

  But, yeah, so the, the video that we just saw, a drone that, that identifies a target and flies into it.

• 21:19Craig S. Smith?

  How is the, is the human watching what the drone is seeing and can abort the attack? If it feels like it's making a, a mistake?

• 21:29Speaker 00

  Of course, a, a, a human could. Also, we can take this a step further. Let's say, rather than trying to blow up the tank entirely and killing the tank operator, we just want to hit the track of the tank and just disable it. We can do things like that as well. Just like with moving, moving trucks on convoys. Why don't we just disable the truck now the convoy is a sitting duck? We don't have to kill the enemy, we can actually just neutralize their supply chain. And we can get smarter and, and, and control the collateral damage and the chaos.

• 22:02Craig S. Smith?

  Oh, that's interesting. So that's, that's part of what you're doing with the, with the vision system.

• 22:08Speaker 00

  That's right.

• 22:09Craig S. Smith?

  Being able to identify components or, or areas on a, on a vehicle.

• 22:17Speaker 00

  That's right. And, and I think a lot of this too is, is the, the economics of warfare. If you look at the invention, the Barrett 50 cal, it's a, it's a, it's a single man operated 50 caliber rifle, but it was really intended as an anti-material rifle to take down radars and grounded aircrafts. And it's a, it's a $10 bullet and you can disable a multimillion dollar aircraft. So a lot of this of warfare too is going to be, you know, economics based. These consumer grade drones are very cheap, but they can also take out very expensive assets. And then, of course, you know, that'll help you win the fight as you bankrupt Russia.

• 22:56Craig S. Smith?

  Yeah, the, how does, how does this, is this going to affect

• 23:07Craig S. Smith?

  air warfare or, or surface warfare on the, on the ocean?

• 23:14Speaker 00

  You know, I think eventually, human pilots will be a thing of the past. I think AI will, will become much better at flying and much better at dogfighting capabilities. And I think humans, again, will be in the loop of sort of, again, orchestrating and quarterbacking. But the days of, you know, the Top Gun style dogfights, I think those will, those will come to an end.

• 23:34Craig S. Smith?

  Yeah. And how, how quickly do you see this moving? Because we certainly saw, I mean, I've said this on, probably to you and, and to others on the podcast, that at the beginning of the Russian invasion, there was that 40 mile line of armor headed toward Kiev.

• 23:56Speaker 00

  Yeah.

• 23:58Craig S. Smith?

  And had there been, you know, enough drones that they would have taken out that whole line. And that was just, what, two years ago? I mean, it's, and now drones are ubiquitous on the battlefield. So it's moving very quickly. How quickly do you think? And another thing is Mark Milley, I heard him at a conference and then he said, you know, the aircraft carrier that, that they're working or building today is going to be obsolete by the time

• 24:38Craig S. Smith?

  it's in the water because you're not going to use aircraft carriers. Everything is going to be,

• 24:46Craig S. Smith?

  you know, unmanned vehicles and that sort of thing. So yeah, how, how quickly do you see this moving into actual battlefield situation?

• 24:57Speaker 00

  I think this inquire, the, the, the entire transformation process, I think, will be less than a decade. I think in in ten years we'll have drones like, I mean, we already have drones today that can approach supersonic speeds. But I mean, to your point about completely unmanned, doing everything, uh, yeah, I I'd say we're we're less than a decade out. Um, just like A, just like the Jarvis experience here, Iron Man, that technology is nearly here and we're building that. Um, that's that's also less than ten years out. Now, when people are saying AGI will take over the world in in twenty twenty eight, I think they're they're insane. And some people we know have said that. Um, I I think AGI is is at least twenty, if not thirty years away. So it's a long ways away.

• 25:42Speaker 00

  But I think AI will become very sophisticated, where it'll resemble reasoning in in AGI, but it's not quite there. Because again, humans will always be in the loop. And so again, that's that's that's our philosophy. Is rather than one big mega model controlling everything, it's going to be swarms of agents. And and these different agents have different core commonities, just like humans do. But again, the the agents will always work for us. And that's the point of it, like, just like Jarvis always works for Tony Stark. He never goes rogue. He sticks with Tony and that's sort of how we envision AI. It sticks with you.

• 26:15Craig S. Smith?

  Yeah, um, the, um.

• 26:20Craig S. Smith?

  Uh, hang on a second, there was a question I was gonna ask, uh.

• 26:26Craig S. Smith?

  Yeah, do you follow what China's doing at all? Because certainly they've been very active, particularly in the drone space.

• 26:34Speaker 00

  Yeah, you know, I think what what's sort of a sobering moment is us losing the DJI protocol manufacturing to China. And so we lost that and we we can't lose the AI race to China. And China's very formidable. Their Alibaba models out of China, the coin series are are phenomenal.

• 26:55Speaker 00

  And so are they ahead of us? You could argue they are, but just the fact that there it's even an argument is concerning. Like we should be way ahead of China and we're not. So yeah, I think we need to, we need to put our foot on the gas. We need to take this seriously. And we need to really ensure that folks that are building AI uphold national security and that protect Western principles and values. Because again, how do we make AI think like an American? How do we make AI culturally aware? If you take a model and then translate it to Hebrew, that doesn't mean it's thinking like an Israeli. It's just a translation. So like, as AI gets more advanced, it'll learn your, it'll learn our culture. And it's important

• 27:38Speaker 00

  that we, that we build AI that magnifies our culture and that protects it.

• 27:42Craig S. Smith?

  Yeah. I mean, I have to ask on this, uh, um, computer vision enabled drone, um,

• 27:53Craig S. Smith?

  could it, uh, be, uh, deployed with facial recognition technology?

• 28:02Speaker 00

  It could be, but then there's the argument of it's unethical to capture people's faces and store it. Um, but yes, you, the technology is there and you could do it. Absolutely. I know China's doing that. I know with GDPR, they're, they're trying not to do that. So if you want to track a human, you don't need to, you wouldn't even need to track their face. You can actually analyze their gait, the way they walk. I walk differently than you do. So my gait can be a unique identifier, even though you're not looking at my face. So things like that is how you get creative and have a workaround.

• 28:37Craig S. Smith?

  Yeah. Uh, I just remember the, you know, the hunt for Osama bin Laden. Um, if there had been, uh, face recognition enabled drones, uh, that, that whole thing must, might have ended much earlier. Um, the, uh, you, you mentioned at the very beginning, uh, the seven seconds, um, that a warfighter has from the time that it's identified by a drone or,

• 29:14Craig S. Smith?

  When it's a drone locks on to the, on to the human.

• 29:19Speaker 00

  Yeah, seven, seven minutes when it.

• 29:21Craig S. Smith?

  Seven minutes, yeah, I'm sorry.

• 29:23Speaker 00

  When it sees you to when you need to get covering concealment. And it used to be much longer. I mean, maybe it's seven, eight, I don't know, ten minutes max. It's very, but it's very, the window's very short. But the point is, it used to be hours. And now that, you know, drones are ubiquitous now and getting more dangerous. It's now shrunken down to seven minutes around.

• 29:44Craig S. Smith?

  Yeah. Is there anything on the, uh, defense side that, uh, that you guys have looked at, uh, that, uh, soldiers on the ground can use when they're identified by, by a drone?

• 30:00Speaker 00

  You know that, I, I, I can't speak to that, but what we're building. If, if, let's say you are in the field, you know, and you have, you have a drone, call it, you know, a, a FAW that has 3D printing, we can quickly assemble the drones and zip them out there. Um, but I, I would imagine fleets and swarms of these, of these cheaper drones. That if I am seen, we can just deploy them right away in my AO. And then if a drone's coming at me, these suicide drones will then fly into it. Um, and so that, that really to me is the solution. I know that there are other companies that are using AI to actually operate a machine gun. And a machine gun will then shoot the, shoot the drones down.

• 30:40Speaker 00

  But problem with that is with, with ballistics, your, your MOA, your, your minute of angle is, is off. So if I'm shooting something a thousand meters out and it's a, it's a, it's a five, five, six round, there's a variance of where that bullet will actually be, even though I'm right on target. Especially with wind age and the air density, you name it. So it's still not very effective. Because you would want a shotgun to shoot a drone, not a, not a machine gun. Or the, I think the best solution will be cheaper suicide drones that'll lock on and, and, you know, and get right to it. And again, and that'll save money and it'll, it won't give away my position. Uh, there's lots of advantages to doing it this way.

• 31:23Speaker 00

  And I think we need more people with actual military experience in defense tech. You know, a lot of these folks are, are amazing technologists and they're, and they're great capitalists, but they don't think like a warfighter. So you can't really build products that, that, that'll augment the warfighter if you don't really know how they think.

• 31:40Craig S. Smith?

  Yeah. Is there any of this technology being developed in Ukraine where you have warfighters that are getting a lot of experience?

• 31:52Speaker 00

  I would, I would imagine so. You brought Eric Schmidt. I believe he's got, he's got some stuff over there, although I can't comment on it directly because I don't know, but I'd be willing to bet that there is. And again, if anyone listens to this, that knows, please reach out because we'd love to help aid the fight in Ukraine with our technology.

• 32:13Craig S. Smith?

  Yeah. Okay. Is there anything I haven't asked that you want to talk about?

• 32:19Speaker 00

  Yeah, there's, there's one last thing too. Weapons detection. We have a very good weapons detection model that detects weapons with, with 95% accuracy. You know, we can also enforce security at, at bases, at FOBs. And this is a technology I think that also needs to become ubiquitous for schools. And so we're going to show you a video. Viewer discretion advised. This is real footage, but you can see the accuracy in the, in the frame rate detection that we can, we can track the pistols.

• 32:53Speaker 00

  And I also think this is the future of, of how do we secure our schools, you know, banks, military bases, you name it.

• 33:07Craig S. Smith?

  And this, and what we're watching are, are two people engaged in a gunfight in surveillance video, but it'll detect, because the, the problem has been over and over again that someone has a phone in their hand and it's, somebody thinks it's a weapon.

• 33:31Craig S. Smith?

  It'll, it'll differentiate.

• 33:34Speaker 00

  That's right.

• 33:36Speaker 00

  Yep.

• 33:36Craig S. Smith?

  Is, are there any examples in the, in the video of, uh, uh, of, uh, phones being, uh, identified as phones in a situation where other people have guns?

• 33:54Speaker 00

  We can, so we can, we can actually build computer vision that'll delineate between a phone and a gun. We didn't for this case, but what I think is interesting is on, on the, on the person's body that was neutralized, there's still a weapon on that body and the camera's picking it up. And so it's important to secure the area, remove all the weapons, things like that. Um, you know, and again, we can put this technology on drones for reconnaissance. We can put it in low power cameras. And again, the vision element needs to augment and, and give us situational awareness on how to do our jobs better. And so I think, I think the, the bigger vision here is how do we connect the physical world, AI, computer vision, with the, call it on-device personalized AI agents? And then how do we bridge that gap?

• 34:39Speaker 00

  And, and I think that, that's the future of AI.

• 34:43Craig S. Smith?

  Yeah. Uh, and this, uh, weapon detection system, has that been deployed or is that a research project?

• 34:51Speaker 00

  Not yet. It's, it's research, but it is ready to deploy. Um, and again, we're, we're ready to provide this to schools for no cost. Cause it's the right thing to do and I never want to profit off of a horrible tragedy. But, um, we will be selling this to our, our military partners. But for schools, absolutely. Um, you know, we're, we're happy to give it away.

• 35:13Craig S. Smith?

  Well, this has been fascinating. Is there anything I've missed?

• 35:16Speaker 00

  I would just, you know, lastly, leave the viewers of, you know, the dangers of AI and the biases and the, um, the dangers of tech elites.

• 35:29Speaker 00

  Trying to censor AI with their own virtue signaling that isn't indicative of our culture or the warfighters. And again, there is no right answer, but I think the most important thing is transparency. OpenAI will never show you their training data. Anthropic will never show you their training data. They won't even show you the weights of their models or the codebase. So forget the training data. And then with open models today, no one's really showing you what they trained on. So if you can't understand the training data, you can't understand the biases. So when hallucinations happen, which they can, you know, you can't

• 36:11Speaker 00

  really go back and understand that calculation if you don't know the training data. So it really all comes back to that. And I think that's the biggest problem with AI today. Garbage in, garbage out, bias in, bias out. We saw Google literally whitewashing or changing history and removing white people, which is, it's borderline evil to do that, to change history. It's problematic and we can't be doing that. So it's, and it's fine if you want to do that, and that's what your customer wants, but now show the training data. You know, how did you do, what did you use to create that? And that's, that's, I think, where the, call it the, the part that I just can't ever stand with.

• 36:53Craig S. Smith?

  Yeah. And, and your training data, you were saying it's, it's open?

• 36:59Speaker 00

  That's right. So we're building, we're in the process, or call it phase two, we're constructing a military data set that thinks like a military warrior. So imagine like noble, you know, Greek mythology with military tactics, with leadership books, lots of curriculum that a young military officer would have to read, combined with lots of academia papers for just the volume. So there's lots of large open data sets on Hugging Face. But then, taking a step further by making it culturally aware. So with a, with a military data set trained off of Air Force, Navy, Army, Marines, all of that public domain. Other than just scraping all of the internet, we're scraping public domain that's akin to the military.

• 37:45Speaker 00

  Combining it with an academic dataset that we can commercially use. Which is right on the cutting face, then you fuse that together. And then you create your small language models. And then you create your LoRa adapters, and then you have your RAG pipeline. And then now, what we do is we create something called function calling. Which that is the start of agents. That's the start of a model doing something for you. So imagine if you say, alright. Summarize this email, send the email out for me, and then schedule a calendar invite with the stakeholder. AI will now do that via agentic workflows. And then I can actually tell it like a human conversation via natural language processing.

• 38:26Speaker 00

  And that's the future of AI, you know, right on the horizon. We have this technology ready today. So we're excited to showcase this at CES in Vegas with our partner Intel. And hopefully more. And, you know, thank you for the time, and I appreciate the platform to talk about our vision and why we're aligned with American principles.