The Drone Company Quietly Taking Over Delivery

27:44And I think they're going to hear a lot more from you guys in the next, I guess, you know, nine to 21 months, kind of through the end of this year, through the end of next year. So if people want to learn more, what's the website? And do you want to shout out a job you're looking to fill here in the US that the right person might to be listening to so the website is mana.aero aero and it's a fun website and then the job is everything but mechanical engineering key electrical engineering software obviously in the usa particularly head of manufacturing so that's the hard one to fill so somebody to lead our manufacturing efforts over there um everything else is all engineering stuff all right and i do really want to thank you for your commitment to when you open your first northeast hub it'll be in Providence, Rhode Island.

28:30You heard it here first, folks. It's a contract. It's the deal. Yeah. No, Bobby, an absolute treat. And I hope everything goes as well as you think, because I'm very lazy and like to stay home. So thank you very much for making my life. Thanks very much. Today, we're going to war. No, we're not going to get into a tank. We're not going to get into a ship. I'm not going to fly off an aircraft carrier. Instead, we're going to talk about the use of drones in war and where they do well and where they don't do well. One issue we've seen across the Ukrainian battlefield after the Russian invasion is that there's a lot GPS jamming going on.

29:00That means if you depended on using GPS to fly your drones to where they're supposed to go, you might have a bad time. Drones sometimes even go off course landing in the wrong country, which is a big problem in international relations. But one thing that we can do is improve the ability of drones to fly without GPS. And that brings us to today's guest, which is a company called Theseus. They are building a system that allows drones to fly without needing satellites to tell them where they are at. Apparently, it uses Google Maps and a camera. Sounds very simple. to tell us more about it. Ian Laffey is here.

29:29Ian, how you doing? I'm doing great, Alex. Thanks for having me. Oh, dude, an absolute pleasure. And normally it doesn't matter where our guests are calling in from because startups are a global game and people travel. But tell me where you're at and why, if you can. Yeah, I'm here in Kiev. I've been here for a couple of weeks and will be here for several more. Just flying our system with a bunch of different people, partnering with drone companies and doing fun stuff. Big cloud providers may offer you cheap compute, but you'll end up paying the difference in engineering costs and hiring extra developers.

30:05You don't want to waste time configuring virtual networks. None of us do. Or your access policies. You want your team building your product. So it's time to look at Render. Render is the all-in-one cloud platform for developers that allows you to deploy, scale, and secure your apps and agents with zero ops. Most cloud platforms ask you to split your focus between product and infrastructure, or they force you into platform constraints. You know you're going to grow in six months. But just connect your GitHub repo to Render and you are live. Web services, cron jobs, the whole stack in one platform.

30:41It's time to find out why 5 million developers are already using Render. 5 million. Go to render.com slash twist and apply for the Render Startup Program. You'll get anywhere from$500 to$100 ,000 in free credits, depending on your stage and who your backers are. That's Render.com slash twist. We're going to talk about what you have built and why, but can you just give me the snapshot of the drone industry in Ukraine? Clearly at the forefront of drone technology, now helping countries around the Middle East defend themselves. But what's it like to be in the mix? I think the most interesting thing about the drone industry here is how competitive it is, how much of a free market it is.

31:22So I think when you look at defense in general in the US, and the way that we procure things, like we are at war now, but the wars that we are at and the wars that we go into are much different from the war here. And here you have like a front line, you have a place where thousands and thousands and thousands of people are fighting every day. And the technology needs to work there. And it's very easy to get it out there. And so I think that that really lends itself to, you know, you hear a lot about scrappy startups, but also just like much faster iteration cycles, because you can take something and actually know if it works on the order of like days instead of, you know, many, many, many months to years.

32:07You're not doing as much testing in the lab versus deploying on the field. So you have a pretty quick information feedback loop on how the technology works and what's doing well and what's not doing well. I guess that's a real benefit to the improvement of drone technology. Terrible circumstances, but at least we're making some progress. how fast is the drone versus drone technology race advancing? Because I know that also, of course, Russia uses a lot of drones and Ukraine uses a lot of drones. So is it something where you can kind of see one side pull ahead and the other side catch up and back and forth like that?

32:41Yeah, I think the big advantage that Ukraine has is Starlink. So they have access to Starlink and Russia doesn't have access to Starlink. um you're seeing more ai drones coming out of ukraine than you're seeing out of russia um i think russia clearly has an advantage in the call it like lucas type you know shahed type drones they produce like hundreds of them every day um and yeah i mean it's it's definitely it's definitely constantly changing. Um, it, it, it seems like a situation where there's, there, there are no winners, right? There's only losers. Um, yeah, it, it, it, it doesn't seem like anybody is kind of breaking away.

33:33Um, as, as far as, as that goes. So Starlink is a great segue into kind of what your company does. Now, when I think about connectivity, I'm familiar with how GPS works. And I know that there's other GPS equivalents from other nations and blocks around the world. And we also now have for the first time really in human history, functional satellite internet across a large amount of the planet. So tell me about the connectivity issues that drones are having in Ukraine, and then walk me from there into how your technology plugs the resulting gaps. The interesting thing when you look at Starlink, right, is it's A, it's very modern.

34:11It's got these panels and they're very directional. And that's not how GPS works at all. You have these geostationary satellites that have been around for decades and they're very, very low power. And they're also in like known frequencies. And so what that means is it's very, very easy to basically shout over them with a different radio and then you end up jamming them. And so what you see here is jamming across the whole spectrum. You know, there's a radio spectrum and you can cover different parts of it with different amounts of power. And GPS is kind of just a band where if you let it go without being jammed, it's just so devastating in terms of precision guided munitions and your ability to collect information and ISR that it's basically an absolute necessity to jam GPS.

35:03And so you see that just universally jammed in addition to communication links being jammed, different links in different places at different times. And so what we do is we kind of circumvent the whole problem of radio, like you said, at the beginning or maybe in the preamble before we started. Like it's a camera and there's satellite maps on board and it tells the drone where it is. And we've spent a lot of time training an algorithm that's able to look at very old, very out of date, very bad satellite maps and very bad camera imagery and still figure out where it is given all of those circumstances.

35:43So the way that I understand this, Ian, is that if you have a topographical map and a camera, you can essentially look out through the camera, see what you see, compare it to the map, and then get a reasonable understanding of where you are. But to me, that does not sound incredibly precise. Now, with GPS, when I'm in New York City, it can tell which corner at an intersection that I'm at. So it's incredibly precise and quick. So how much fidelity do we lose by going to your system? To be clear, I understand that it's the only one that works. So I'm not trying to be unkind, but I'm just curious about how well we can do compared to the GPS that people listening to this probably already know and understand.

36:27It's a great question. GPS is amazing. And I think what's really interesting about this is you've had these kinds of solutions for decades. If you look at something like a Tomahawk missile, right, like that has a laser rangefinder and a really good topo map and it can figure out where it is. Really, what we see ourselves as doing is like taking that from like, yeah, government level maps and a really nice rangefinder onto, you know, like a$25 camera off Amazon and very low resolution maps. And so to answer your question, yeah, the answer is not very precise. So we advertise 30 meter median accuracy.

37:15And that is enough to get you from A to B, but it's not enough to tell you where you are on a street corner. But that's kind of also something that's pretty important to us. Once you start getting into higher levels of precision, there's a lot more considerations around that kind of technology. So 30 meters sounds like a lot, but actually I don't think it is because, and I don't mean to be kind of crass about this, but some drones go boom and I wouldn't want to be 30 meters away from one of them doing so. So I presume there is a margin for error in the marketing. Yeah. The, the, the going boom is not, you know, it's, it's not really like the core focus for us.

38:01What we're focused on is getting a drone from A to B over very long ranges and over very complicated environments. So if you need to fly a drone for 600 kilometers, you know, in a straight line over fields and still know where you are, we can help you with that. But, you know, that kind of like last mile targeting and like strike capability has just not been a focus for us. So I think then that we should better understand what that use case is you're describing. When do I need to send a drone in a straight line for six hours? And to be clear, I'm exposing my lack of operational knowledge of drone warfare here, but that's not a circumstance that I kind of expected unless you're going deep into an enemy country, for example.

38:51You know, back to the topic of radios, right, is like to maintain communication with a drone is difficult in a jammed environment. And so if you want to go and you want to take pictures of things and fly around and stream video back or go and record data over someplace without getting jammed and be able to go complete that mission and then come home or go fly that mission and then stream data back over Starlink the whole time. These are things that we help with. And I would just say that is generally like this stuff that we like to talk about and focus on is kind of those capabilities. My understanding is that Starlink is actually pretty difficult to jam.

39:39Is that correct, Ian? Yeah, it's just like a physics problem. They have a highly directional antenna and it makes it much more difficult to jam that kind of phased array. So in time, do you think that we get smaller and smaller satellite internet receivers, a la Starlink or whatever from Blue Origin or from the E or whatever? And does that dramatically change the state of drone warfare? Or do you think that we're still going to need GPS replacements without kind of real-time satellite connectivity to ensure that we can keep doing the drone operations that we need to get done? I think you said kind of two things there.

40:22And the first, definitely, like you're just going to see, you know, you're going to see Starlink direct to sell. You're going to see like a tiny little chip that connects your phone, your phone or your drone to a satellite from anywhere in the world where SpaceX lets you do that. The thing that I would say there is that, you know, communications and positioning are kind of two different things. So like you think about your iPhone, you have a LTE network that lets you call people and, you know, look at Twitter. and then you have the GPS, which lets you, you know, you combine that with the LTE to then use Google Maps or to find where you are in the woods.

41:01And so, you know, it's kind of two separate things. The other thing I'd say that's super interesting with Starlink, and I think this was something that we always viewed as like a risk for our business, but it's been interesting to see them move in the opposite direction is Starlink actually, because it's satellites flying around, they know where the satellites are, like you can position off of Starlink. And so a lot of people in Ukraine, as well as in the US had begun to use Starlink panels and use a positioning feature that's already in them to fly their drones. And there, you know, there's some pros and cons to that over vision.

41:35But I think this goes live in about seven days, it may be like nine days, but Starlink has basically just like completely blocked that capability or made it like so inaccurate that, you know, it's, it's maybe like 50 kilometers precise unless, unless you pay them a bunch of money for like military Starlink, uh, contract. And so, um, you know, that that's just not really the domain that we operate in. Um, for me, like the point of a drone is, is that it's cheap. Um, and we in the U S and other places have like really great technology. Like I think you're always going to need a great aircraft. But there's a certain value that you get to making something just like dirt or cheap and making millions of them.

42:24Well, I mean, that's why using a camera and maps is incredibly cheap. As you said, you can buy some of that stuff off of Amazon. So tell me about the company's product itself and the fact that it can work on, I think you guys say pretty much any drone. I would love to better understand how it attaches. And also, what does it cost to get one of these onto a drone today? Yeah. So we, um, right now, uh, what we have is, uh, so we started off, I guess I'll, I'll, I'll back up is that we started off with, with this box. Um, and you put this box onto the bottom of a drone and it, it plugs in and it pretends to be GPS.

43:00Um, and there's actually like a bunch of people that do, uh, similar stuff that there weren't really when we when we started but there's a bunch um and what we found with that is that it kind of sucks for a bunch of technical reasons i won't get into and so what we realized is that um you can actually just pull the software out of that and design it in such a way that a you can run it on basically any computer so you don't need a gpu you don't need anything fancy you can run it on like a$65 Raspberry Pi. And then also building in just a variety of cameras. So thermal cameras, electro-optical, or just like regular cameras and supporting all of those things out of the box.

43:42And so for us, that's kind of been the evolution of a product is like, we still sell that as physical hardware, but then also offering that to customers who they themselves have really developed supply chains. They already have the compute on board. They have the ability to go and buy these systems. And it's like, you know, very open book. I mean, I'll even say it here on the podcast, right? It's like, it's a camera and a Raspberry Pi and an SD card, right? There's no secret sauce. Don't give away the secret sauce. I mean, I think I could probably actually put those together over a weekend with enough Claude credits, you know?

44:20I think you could definitely put it together. Yeah. It's the hardware is it basically needs to be a commodity. And honestly, like the whole thing needs to commodify because these things just need to get cheaper and cheaper and cheaper. And this has been kind of something that I've been thinking about a lot is like, you know, position, position as a product, it's like, you don't need to build a differentiated positioning product, you need to build a really good positioning product and deliver it at a lower cost basis and a higher scale. This reminds me a little bit of like the automotive industry, because one thing that someone told me who works in robotics is that, you know, if you make a component for automotive, they're going to grind you down on cost.

45:04And they're also going to demand that it works for like 200 ,000 miles being bounced around in a hot temperature. So they just want simple, high-performance, cheap. And it sounds like that kind of maps to what we need for our drones. But Ian, it sounds like then the company Theseus has become much more a software company than a combined hardware and software company. At the end of the day, I see us as a reliability company. Our job is to solve the problem of autonomy. So since we've started, our goal has been to keep good guys from getting killed. And what that looks like to me is having less people on the battlefield where they can get killed.

45:47And the way that you do that is you make systems more autonomous or you make the systems that are piloted by people able to be piloted from farther away from the dangerous places. And so in order to do that, you need to make those systems reliable because really the job of an operator in a combat zone is reliability. Humans are very good at solving, you know, problems that crop up on the fly, we have very high common sense. And so we kind of fill in the gaps in reliability in systems. And so how I see our company is, is that we take these systems that are pretty unreliable. And, you know, you could say like janky or hacky, and we make them more reliable.

46:29And so what that looks like today is, you know, the most, most foundational aspect of autonomy. Like you can look at a drone that was made in like 2003 and before jamming that had a higher level of autonomy to any drone that you're going to see in Ukraine flying up until maybe the last year or so. And so by just delivering the position to the drone, you're increasing the reliability of it and you're increasing that autonomy level back up to something that's on par. And then we see that as just like extending out farther and farther from there. So essentially, we got really far with drone technology when GPS worked, just generally speaking.

47:12And now that we are in a post GPS era for warfare drones, we've had to essentially rebuild to get back to the level of autonomy that we had. Okay. Well, that's difficult, old, but I'm glad we've gotten there. What can you tell me about progress in the Theseus technology? Because you guys have been at this now for a couple of years. You're in Ukraine, you're doing, I presume, testing of some sort, but how quickly does the technology improve? And is it improving along a accuracy axis or more of a cost and reliability axis? With accuracy, there's like, you know, just regulatory considerations. So that's not really like a priority for us.

47:52We actually don't want to make our system more accurate right now. That being said, yeah, just reliability and cost. I think you hit the nail on the head there. And it's really just a matter of like, I think even myself, like I'll just speak for myself here. Like it's it's very, very, very, very, very, very different to like field a piece of technology in a war zone versus not in a war zone. And so I think for us, the first half of our company's life has just been getting our product to even work well enough that it makes sense for somebody to use it. And then the past year has been, oh, okay, we know that this works.

48:40We know that it works well? How do we get this to scale 1 million? And what are all the roadblocks that we need to knock down in order to do that? How many drones a year does Ukraine need? And what do you think should be the capacity of the United States to manufacture drones? Because I know there's been a lot of talk about re-industrialization of the United States, building our manufacturing capacity back up. But it's hard for me to tell where I sit if that's a lot of people cosplaying in Bob the Builder hats, or if they're actually making real progress towards shovels in the ground? I love all drone companies.

49:15They are our customers. So they're good things to say only. So Ukraine right now is at like, I don't know, six to eight million drones a year. The US, we have this historic drone dominance program. That is, I think it's 300 ,000 drones over the next two years. Um, I I've talked to, um, you know, the, the guy who runs that program and what they've said, which I think is really good is that the, the focus of that program is, um, yeah, it's to get the one way attack drones. That's very important, but it's also, uh, kind of a shock to the supply chain to incentivize American companies to bring components out of China.

50:03Um, and I, I think that like everybody talks about this, so I'll, I'll try not to harp on it too much, but it is really, really surprising the level to which everything is made in China. And I'll just take one example, which is like a camera. You can get a sensor that's made in Japan or Korea pretty easily, like the actual array of silicon that takes in the light. To get the lens made outside of China is more difficult. And then to get, if you as like just a small business, not like a larger company wanted to get the PCB that goes on the back of that sensor and get that made outside of Shenzhen, you basically can't do that without paying like four or five, six times as much.

50:52And your options are just like so, so, so limited. And I think you basically just see that across every level of the stack. And it's basically all these little load bearing points where you rip one thing out and you go down like two levels of the supply chain and then you realize, oh, actually, this one very niche thing is only made in China. Are American drone companies, both startup and from maybe even the primes, are they doing the work to reconstitute those supply lines here in the States or for the foreseeable future, in are we not going to be able to do enough in nation manufacturing to kind of like supply our own drone components i'm scared now i've been really impressed by what like the the top american drone companies have have been doing and how creative they get and and also like i think yeah time is a factor and and having demand is a factor but you also realize like Like none of this, none of this stuff is that hard individually.

52:00Like in aggregate, it becomes basically impossible. But like that PCB example, it's like, okay, well, if you have this, if you have this demand, like you can put together PCBs in lots of places. You just have to get enough people together that they go, go to somebody in a different country and they say, hey, we need this. The key thing that we need to figure out in our country is like, how do you have a real commercial industry around these microelectronics because you you look at like the american industry in you know the 40s 50s it's like this heavy industry you know you mentioned the automobile industry um that lends itself to these like very large systems and we have a culture that lends itself to that we have a culture also that lends itself to like information technology right like Like we were the country that first invented the filing cabinet just because we love information so much and like storing it.

52:57So then of course, when a computer comes out, like we just have this domestic demand for computers. But then you look at China and they've just been making all the world's stuff for so long that they have all of these smaller component manufacturers that serve a very large domestic market. And so now that we're seeing like war or defense change to bias towards small electronic components, and we're having to ask ourselves like, oh, this has significantly more of an impact on the battlefield than we had thought five to 10 years ago, just cheap, small electronic components. like what can we do in the U S such that we are, you know, we have the domestic demand at scale for cheap, small electronic components.

53:43Do you know, um, that, that, uh, that project cost plus drugs, it's essentially like, I think it's maybe it's Mark Cuban who's working on this. They essentially just make generic drugs in the U S for a very low price. I feel like we need some billionaire to be like, I'm going to make the PCB base company and I'm going to make PCBs at cost. And we're going to make no money but if you need springs pcbs you know carbon screws whatever the fuck we will make it and then just like just do it with the existing capital base because this is not impossible it's definitely not and i i think defense is like a great reason to focus on i just i do think i i don't know i mean i get very scared because i i just don't think like i i don't think that there's that much of an incentive in our current commercial industry like without a ton of protectionism for us to move like back towards manufacturing.

54:38Like there's, there's so many regulations against making stuff. So few people, like we, we have like such a high labor costs that it's like, you know, and people talk about automation. I think that's like one way out, but it's like, you know, who, who is going to be the person that is putting together these, these tiny little components all day for$2 an hour. And how do we even compete against that. And so I don't really have the answers to that, but it's definitely a huge problem because I think we're just kind of waking up to the fact that a lot of things that we have not invested in turn out to be very, very important in a global geopolitical arms race of sorts.

55:30Yeah. This is why I think even us, you know, staunch, no industrial policy folks are starting to go, well, what if we had a little bit of industrial policy? Just, just a dusting on top, you know, because capitalism is, is amazing and such a great engine, but it does not always foresee national security needs. Well, we're also highly regulated, right? Like we, we have so we, we have a minimum wage, we have environmental protections. Like I'm not saying get rid of the minimum wage and like child labor and dump things into rivers. But like, you know, you don't necessarily need more laws. Like a lot of it can be solved also by having less laws.

56:10But I don't think that there's a lot of demands from our population for that. Yeah. Also, I would say like I think we think about this. It's in vogue in like Silicon Valley and like among people that that make things to be talking about these issues. But like maybe this is a little pessimistic, but I just don't think that the average American like particularly cares that much about about a lot of these issues. The average person only cares about what they've recently seen on television and what their friend told them. Like we're not talking about. I don't know. I don't know if that brings me down.

56:49I mean, if if we only do things that the average person cares about, we're not going to do much, but we will watch a lot of American Idol. I guess I'm just saying I don't know if you're going to see like broad popular support for like, you know, deregulating critical industries, because I think if anything, we're seeing like an increase in demand for regulation of critical industries like nuclear. again. I don't know. I think maybe people just spend a lot of time on Twitter. Everybody wants to deregulate nuclear, which I think is good, build more nuclear reactors. That's kind of a new thing. So that's thumbs up.

57:26But you also see we want to regulate ourselves out of AI. And I think it's also like, yeah, I think America just has a lot of... We have a hard hand that we've been dealt and I think we'll figure it out. But, you know, the thing that keeps me up at night is like, we, we allow ourselves to, uh, be exposed to all this information from every country in the world comes onto our internet and speaks to us in English and, and convinces our citizens of different things. Um, and like, you know, I spent some time looking at like, what would it be like to like do propaganda on like behind the Chinese firewall?

58:06Like if I wanted to run, it's, it's impossible. Like you just can't do it. Um, but it's like well-documented and very easy, uh, to, to influence the minds of, of Americans. And so I think that really works to our detriment. I could talk to you about that issue and that set of issues for about seven years, but for the sake of this interview, I'm going to loop us back to Theseus. This is not sadly the Ian and Alex talk about the future of industrial policy hour. Um, so before I let you go, I'm really curious about your time at YC. You guys were, I believe, uh, the company said in a release, I think the first defense tech company to go through the famed accelerator.

58:40So just really quickly, how was that? And do you think that it would help the issues we're describing if there were like another like 50 theses equivalent going through American accelerators to jumpstart these processes? Yeah, I mean, I think it was great. I really have never experienced like a higher density of really talented, intelligent, like kind, uh, motivated people, um, then, then in YC. Um, and I think we had like, you know, Aaron is a really great group partner, Aaron Epstein, and just like, it was a really great experience. Um, I think they've started to invest like way, way, way more, um, into defense.

59:23I have heard a thing that like with accelerators, it typically just follow what the founders are interested in. So I think there's just been an uptick. Um, you know, for example, our company started out of a hackathon. Like I didn't wake up one day and, and, and be like, Oh, I want to, I want to start this company. It's just, there's, there's so many problems. Like even to the conversation we were just having, like, I think you can look at the world and you can see these problems and, and feel like, Oh, there's nothing I can do. Or you can just pick an issue and start working as, as hard as you can towards it.

59:57And so luckily I think you see a lot of companies, um, you know, coming out of the US and Silicon Valley now that are focused on kind of these key aspects of working in the national interest. I think it's already started happening. Do I think that there could be more of it? Yeah. But then also to the thing I was just saying, like, maybe also like commercially focused, not just defense focused, because we're not like, we're not Ukraine and we have a different problem set than, than they do. And we're not actively being invaded. Well, we are currently in a different region dealing with drones. So, you know, I don't think, look, no matter what you can say about America, we do get busy.

1:00:48You know, we have a history of being a busy nation. So I I think there's a lot of places we get to play this technology. Ian, listen, thank you so much. What's the website? And I do hope that tonight you get some more sleep and that Russia does not bomb Kiev as much so that way you can actually sleep. Yeah, thank you. I appreciate it. It's theseus.us, T-H-E-S-E-U-S.us. Awesome. And I don't get to say this too often because I try to stay pretty much impartial with founders, but I hope you kick maximum ass and I hope the technology goes even better than you expect because I would like Ukraine to win and soon.

1:01:20Thank you for coming on in. Thanks. I appreciate it. Take care.