The Vertical Space

#95 Serhii Kupriienko, Swarmer: Building swarms in Ukraine’s drone war

• Luka T • Episode 95

Share episode

Drone swarms are no longer science fiction - they're becoming one of the most disruptive technologies on the modern battlefield. In this episode, we speak with Serhii Kupriienko, founder and CEO of Swarmer, a Ukrainian startup building autonomous swarming capabilities for drones operating in the harshest conditions: GPS-denied, communications-jammed, and under fire.

We dive deep into how drone warfare has evolved in Ukraine, the maturation - and limitations - of FPV drone tech, and what true “swarming” really means. Serhii shares lessons from the frontlines of defense innovation, explains why adoption of new technologies can be both urgent and constrained, and lays out the challenges of building autonomous, coordinated robotic systems in real-world environments. We also touch on counter-UAS and the broader Ukrainian startup ecosystem. 

Serhii:

Swarming is not automation. Swarming is not switching the one operator between many drones. Swarming is through self-collaboration and self-communication by drones. So they work together to achieve the goals that humans set or another drone set probably in some future. And this is what swarming is for us, how we define it. Not about like"you do this, you do this, you do this, you do this, you have five minutes". That's not swarming, that's just, micromanagement.

Luka:

Drone swarming is one of those buzzwords we've all heard, especially in the context of the war in Ukraine. But despite the hype, it remains a hard problem and for good reason. Making drones communicate and collaborate to achieve a tactical effect, particularly in environments without GPS and with heavy jamming, requires some serious engineering. And today we're joined by someone who's deep in this fight: Serhii Kupriienko founder and CEO of Swarmer, a cutting edge defense tech startup building autonomous drone swarm capabilities. In this conversation, we explore how drones have evolved over more than three years of full scale war in Ukraine, from early improvisations to the growing maturity of FPV drones and the complex tech stacks that now power them. We get into GNSS denied navigation, automated terminal guidance, and how even today many of these technologies remain brittle or half baked in practice. Naturally, we spend much of the episode on swarming: what it is, what people get wrong about it, and why it's not just about flying in formation or controlling multiple drones from a single operator. Instead, swarming is a behavior. It's about groups of drones autonomously coordinating and adapting to achieve a shared mission objective. We also talk about Ukraine's unique, innovation dynamic: rapid necessity driven invention on one hand and on the other, a battlefield environment that often favors familiar tools over unproven ones, no matter how advanced. Finally, we discussed the biggest hurdles to developing and deploying swarming systems, the future of counter UAS technology, and where the Ukrainian startup ecosystem goes from here. A bit more about Serhii, before founding Swarmer, he led AI efforts at Ring, the smart home security company, acquired by Amazon. He's a Stanford alum and also studied the ethics of AI at the London School of Economics. Today, he brings that rare mix of Silicon Valley engineering and thoughtful systems design to one of the most difficult and urgent problems in defense tech. Enjoy the conversation with Serhii after a brief sponsor message. Well, Serhii Kupriienko welcome to The Vertical Space.

Serhii:

Oh, thank you for hosting me. Pleasure to be here with you guys.

Luka:

too. We're looking forward to this conversation and we normally ask the first question if there is anything that very few in the industry agree with you on.

Serhii:

Yeah, I'd say this is a very good question and the point is, swarming is not about flying in formations, like literally, because, that was my, my very big surprise when you saying, okay, swarming is about efficient, management of different assets. But when you show up into the generals and to the industry primes, they say. What kind of formations do you fly in? We say we don't like. No, this is swarm You should. And end of the day, I just gave up and say to my R&D team, guys, can you please implement some formations because I just don't want to explain it anymore. Okay. Formations. Formations. But this is the one of the biggest things.

Luka:

So in other words, you're saying swarming doesn't necessarily mean the proximity of one drone to another, but it is about how they are, deployed in a collaborative, fashion to achieve a certain effect.

Serhii:

Absolutely. Uh, kind of, where did it all came from is because, birds, fly in swarms work in six. But that's because they use visual, connection to each other. And this is how they perceive and, send the signals to each other with wings, with the body, uh, position. For drones that's not the point. And the second point is when you have many assets in the same area. You have the small interval and high regions of this assets when you want to have like five different strikes in a row. But again, this is not necessarily swarm. You can distribute it across the area from different sites. It always, should be defined by the, if you're talking here in defense, by the enemy systems, by the location, by the environment, but not just we have a square shaped swarm and we like, we win everything. That's different story.

Peter:

And do people have a misconception about the size of the swarms that we're talking about? I mean, lots of people have seen drone light shows and the like, and are they incorrectly transferring that sort of mental picture over to what we're talking about here?

Serhii:

Yes, like that. So if like, at Swarmer if you're a successful company, you should present something with the sky, like completely dark out of the drones. No, it's not. And, you know, the, thanks for the question about the birds. I think, one of the key differences we bring here, is we are building the system by first principles, how it should be done for computers not mimicking humans. And swarm is somewhere from the nature when, If we try to build the group behavior, we don't have to try and mimic, nature here because computers are way more efficient in different ways, and we should follow the strength of the computers and all the communication systems we have there.

Luka:

Another misconception that I commonly hear with regards to swarming is that people tend to forget that swarming is a behavior and not a a type of flying. Right. Which kind of ties back to your observation, and I am a little bit surprised because, especially people who are, In the armed services in the Air Force, they should know, by observing the evolution of formation flying that initially, you know, aircraft started flying really close to each other because that was the only way to support each other. And it was always about visually checking, the six o'clock of your lead, and mutual support. But you look at the, fifth generation, fourth generation fighter aircraft today, those formations are so-called sensor formations. They're staying in formation to the point where they can have situational awareness by means of sensors. They're not necessarily close to each other the way that aircraft were in, the early days of aviation, or even World War II or, or hell even, you know, later, in conflicts. And so, yeah, it's surprising how people jump to that conclusion, that formation flying and slash swarm implementation in drones necessarily implies flying close to each other.

Serhii:

I think that makes, makes perfect sense. And The very root reason is, look, you have to be engaged in innovation somehow and in military aviation to know this. and who are the people on the front line and in the armed forces who work with drones right now? They just infantry forces, intelligence service, special service like everyone except aviation gate. So they have totally different experience, totally different perspective. They have a different even, understanding of, network centric warfare because for them it's different.

Jim:

I'm also assuming given we're gonna stay on the name Swarmer, you're dealing with the greatest fear

Serhii:

Absolutely. Even two of them. Yeah.

Jim:

So it's a, I would assume, I love the name of the company, but I assume you're first dealing with, the military's greatest fear before you actually explain to them what you do as a company. I.

Serhii:

So the, when I try to explain swarming to the people who are not in, I use the very, very simple sentence every single time. The swarming this is not when you have five drones flying the same direction, basically this is called automation. this is when you have five drones in common, you should first and the rest immediately regrouping, strike you from every single other side. So this is what makes you so exciting and so scary at the same time. and, this is so true and I constantly get asked by, okay, Will it work? Yes. Can we stop it? Yes. how should we do it? This is the big red button. Okay. May I press it? Yes, please do. Will it work? Yes. And, because, yeah, this is AI and this is some like crazy unclear, how to understand AI and the, on the one hand. And on the other hand, especially when you work in Ukraine for the last three years, army developed a lot of practices and basically they get used to the, assets they have. That is a very simple human controlled drones. So as soon as you try add a piece of automation or autonomy, it's totally different approach because now it's, it's not a soldier with a remote control from two kilos of exposure that's flying above your head. That's some crazy self, driven, two kilos of exposure flying above your head. And oh, that makes a lot of things different.

Luka:

Well, we'll get to that but how about we start with, walk us through the arc of how drones have evolved in Ukraine since 2022. And at a high level, I will try to, summarize the first year of the war as the war of, you know, the basic Mavic ISR drone. The second year was about FPV drones, and then the third year perhaps about, air interceptors and some more advanced tactical use. I. Would you agree with that? You know, high level arc and give us a bit more detail of how you've seen it from your perspective and how this arc continues to evolve.

Peter:

And Luka, I think it's important to also inject the scaling that took place over those years. Not just the evolution of use cases, but, there was, obviously tremendous scaling and bottlenecks were encountered, and I think that this story fits into that as well.

Serhii:

Yeah the first drones in Ukraine was applied in 2014 and 2015. Like they very, they very early companies, and the DJI started to emerge then. But I'd say it's pretty similar to the development of the piloted manned aviation, during the Second World War and just before that. So, and what drives me the crazy here is, we tried as a industry a lot to bring way more automation, way more kind, uh, autonomy to the drone. But, 99% of them are still, remotely controlled toys and require from one to five pilots, depending on the class size and, just, just to manage them, just to get them airborne and to control them. But what changed dramatically? It's not a garage made drone anymore. That's, highly industrialized, assembly manufacturer lines for the small ones, even for the big ones like companies like Airlogix have, or, Sky four, for instance. They have decent manufacturing, capabilities, and they do it, they do thousands of drones and this's, highly automated manufacturing. Other than that, in terms of the technologies, we, developed a lot of things, again in terms of material design, in terms of assembly lines, et cetera, et cetera. But in terms of the underlying technology, it's still the same open source, mostly open source or Chinese source flight controller software, and, the development arc is close to the, again, to the aviation. It's small drones, the bigger ones the counter UAV systems, the faster ones to avoid counter UAV and drone carriers that deliver small guided bombs, or in our case, small FPV drones. So that's all kind. Nothing, nothing dramatically new, fortunately, and unfortunately.

Jim:

How has the drone been used in the war over the last couple of years? What's changed? How has it been used? And obviously even the last week, I mean, a, a rather sizable forward, missions into Russia. Explain how it's been used during the war.

Serhii:

I'd say for the last three years, the roles, haven't changed dramatically. You still have the FPV drones there. you still, have ISR drones of different, different classes for different distances. What, improved that the integration, the actual reliability of every single drone and the scale. So, I'd say out of the, all the technologies, one of the most surprising as for me because I wasn't the believer in the fiber optics technology at all. so kind of sorry for that. I'm a bad, bad future, describing guy. But, when you have a unjammable remotely thrown granade with a video camera, you can, control, that changes a lot. If you can throw it precisely for 25 kilometers. and, over the last year, uh uh. thanks are up to the increase of the distance of the drone operations. the gray zone, significantly extended. So when it was 2022 when we started everything, it was like five to 10 kilometers probably. And those was covered by artillery mostly and by, uh, mortars. Right now they're at 25, 20, 25 kilometers from each side. So, and, everything is eliminated by drones. It's way more, way more efficient. And in terms of the mass adoption, way more adopted weapon than artillery uh, that it was in 2022. Do.

Luka:

How do you expect the FPV drone, to evolve, in the coming years? Do you think that this will be the form factor that sticks or will it be changed? Obviously the mission will probably be here for a long time, but is the form factor going to change do you think?

Serhii:

kinda, I've just told that I'm really, sounds like I'm really bad talking about the future. But nevertheless, I think that, I do believe that there will be like two directions of the development. The first one is the drones that will work together, that will be like autonomous launching from containers. So you can really launch a dozen of them, but that will be more on the kind of brigade level or regiment level. So covering like 10 to 25 kilometers of, of the front one because it'll be the very, very connected network of assets. And basically we believe that we will contribute a lot to such a direction. But, on the other, hand, I believe that, the FPV drone it already became, and it'll be there for the long time, just a part of the equipment of the infantry soldier. Because end of the day, whatever you do, you capture the territory only when there is the human, to put a step, in the ground. It means we need infantry there. It means, right now, look, the FPV drone training for the infantry soldier is like another piece of the training. So you have tactical medicine, you have like rifles, you have grenades you have FPV drones, you have like tactics, trenches, everything. That's a little part of the educational process of the training process. I do believe that they will be smaller. So you don't need the, you don't have to have a 25 kilometers range drone for the infantry guy, but if you will have three or five of them for five to 10 kilometers each with a very, very simple control, it's just a precisely guided grenade. It'll help you a lot. The next level will be covered by the, by different organizational level as you just described.

Luka:

Do you see that there are certain features or conops or things about drone technology that the end users, operators on the front lines are consistently asking for that the engineers or product developments and drone companies often overlook or misunderstand or don't deliver.

Serhii:

I think the key point look one of the technologies that failed over the last year was actually that, that the terminal guidance. And, the way how the drone can hit the target automatically. not because, less because of the work of the technology, but way more because many teams started to do this as engineers they paid zero attention to the actual customer need and actual customer experience. And, end of the day, in half a year after many teams and many deployments, army figured out that it's way better to invest$200 to the better connectivity system, to the better communication model than to terminal guidance because, it's less risky. It's way more easy and way more clear and you can just work, work with, with every single drone you used to work for the last few years. Answering your question, I think the most, the most really demanded technology is just throw it away and, it'll hit someone like, because usually when you need drone for defense operation, I mean, when you execute the offensive operation, you can plan it accordingly. You have all the assets you can plan and execute artillery support intelligence support, et cetera, et cetera. But when you, uh, play defense, that's different story. And you have to have something that reacts really fast and as kinda self-confident, as autonomous as it could be. So throwing out the drone and saying, there is someone five kilometers in front of me, just find them and strike them. This is kind of the silver bullet nobody delivered yet. Of reasons first of all, the immaturity of the technology, kinda because the whole computer vision in our world. I worked for this industrial 15 years. The whole computer vision was trained on cats from the internet and from kind of trying to split the cookies from the, from the puppies. But, there is nothing that works on the battlefield. So there were high promises. unfortunately, very, very little deliveries.

Luka:

Can you explain more, the distinction between investing in connectivity versus terminal guidance?

Serhii:

Yeah, every single normal armed forces unit right now has its own like small R&D lab. It's not called R&D lab, but they said a drone department. And they have tons of spare parts, tons of frames of different, they know how to work with radio links. Like what is the difference between ROS of different versions of VTX of whatever, like all the stuff, because they basically assembled, in their location, in their headquarter and send it to the front line because, 90% of the drone you got delivered, as arm forces unit from the central warehouse they just don't work due to different reasons because they're outdated. They, you, you need different frequencies. You need different, like, et cetera, et cetera, et cetera. So they rework everything. and when, you try to add a little bit more expensive, but like way more EW resistant, communication model, that's the same. You just put another model, but everything other, it's still the same tactics, preparation antennas frequencies like you. This is why fiber optics, drones work pretty well. You change virtually nothing. You just have to fly a little bit slower, a little bit more carefully, and it's unjammable, boom. Everything else is completely the same. No changes, no retraining, no answering dumb questions from work, from your, unit. just do it again. With terminal guidance, you have to do a lot of things differently. you have to, you have to get adopted to its limitation. You have to get adopted to. Its like, special way, how to how to take the appropriate angle of attack for the target because it's just a claim that you can, you could do a pixel work and it works. It's not, there yet. And you have to retrain your people. And the added value is very, very kinda, neglectable because it's way easier for you having everything prepared just to put more power to the antenna to find a better place for the radio relay drone or radio relay station and you can strike. it precisely, that's just because balance of the ROI versus efforts and versus unknowns.

Luka:

I'm jumping ahead a little bit, but for what you are working on with swarming, how does that trade off then look differently?

Serhii:

There is one interesting, or surprising, surprising point and surprising customer, how to say it, end user concern. They say what do you mean drones will fly by themselves. You don't need us operators anymore. So they will, took us and put us to trenches? No, we don't want the system. so this kind of bring a steam machine to the, to the land like 300 years ago. On the other point. we kind of to try to jump this gap because we're not adding here, and this is impossible to add incremental, improvement, like a small one because you will always face, oh, we don't need you, we just got a fiber optics based drum, for instance. We are presenting the multiplier here so now you are an operator. You can control many drones at the same time. You still have sense of control. You still can do everything. You still use the same, virtually the same communication systems. you get used to, but now you just stand not one drone and put everything under your fingers. But you have 10 drones at the destination area. If you want, you can jump in and ride the control and do the strike by yourself if you're sure it works. you can just, rely on the automated system. We work pretty carefully with step by step trying to end pretty close to customers trying to. find the fear and manage it. So, and sense of control is one of the biggest fears here.

Jim:

what have been some of the more important tactical lessons about drone employment in Ukraine that other countries have not yet internalized?

Serhii:

Hmm, that's interesting. Okay, lemme let me try to tell it this way. the very, very key difference between having FPV drone is, that 90% of the previous, approach to the warfare is now irrelevant because, imagine you like 25 years ago, and so you have tanks, artillery, EW systems, uh, SIGINT systems, everything but you to hit something, you have to get the, visual on the target and then you can strike it with the precision guided munition, I don't know, from 25 kilometers away or from, or some bomb or some rocket. But you have to have visuals end of the day or GPS connection. Right now, every single soldier for the range of 25 or more kilometers can have the combined ISR and immediate strike capability. So, that makes, and you cannot cover it like with, I dunno, with fences, with like hangers, with anything because, you two soldiers, the first strike, the hangers, the second strikes everything inside. And that makes the whole game different, to the very, very core level work. And how do you protect your assets? How do you, detect the threats? How do you counter them? it's not enough just to put like a high fence, put everything to the concrete hangers and everything will be intact. No way.

Jim:

Yeah, that's really quite fascinating. It's amazing. and, and is it your sense that other countries haven't yet come to grips with that capability?

Serhii:

I think even Ukraine didn't get it to the a hundred percent yet. Because for us, it's just a matter of survival. we just adopt and ingested as it goes. But, the one that interesting, it's not because I'm so smart here, just because I was trying to figure out why the latest strike to Russia, military Air Force base was so kinda loud because everybody was quiet for two days. and, and this is one of the reasons why, because you, it's finally became very, very clear.

Jim:

So when you have a 20 to 25 kilometer, precision viewpoint that drones are providing, what's more likely being acted upon as a result? Are they bringing the munitions down via drone? Are they calling in artillery? Are they calling in it's probably not air support. What's the more likely way that that infantry man is engaging the enemy with that drone 25 kilometer, precision viewpoint? How are they acting on the target? And I'm sure it depends on the target as well.

Serhii:

This depends on, on the target, but again, the key point that you now have combined capability. So as soon as you see the target, you strike it immediately because you have the camera and the explosive by itself. Before that. Before that you had two different assets. You had eyes in the, in the sky, but it was kind of some intelligence big drone, without strike capabilities. You have something on the ground like it, granade launcher Is it mortar? Is it like artillery They don't have eyes. You have to coordinate them together. Now you have combined capabilities. That's very, very different.

Jim:

Talk about combined arms. Yeah, that's impressive.

Peter:

What are the maturation trends going on in the FPV drones themselves in the last 18 months? Because if, if we're going to be networking these drones together and allowing'em to coordinate, there has to be some homogeneity on the software that's running on those drones, on the compute capacity that's on board them, you know, if, if the drone industry began with this really dispersed, you know, rapid high volume manufacturing and you ended up with a lot of different types of drones in the air as a result, what are the trends now towards, enabling them to mature and to work together? What are the things that people are emphasizing and how is that contributing?

Serhii:

Basically this is the problem you're solving in our company, connecting that all together.'cause as soon as you have, like I read on the report two weeks ago, I have to check the source, but for the last year there were 2,500 drone manufacturers emerged for the last year, like 2,500. It means, just because it's really, really easy right now, you have to have virtually nothing to start the drone manufacture. And this is basically the reason why, my advice to anyone who thinks starts in a drone factory in 2025, don't do this. It's useless. And, because you have open source software for the basic stuff, you have, a lot of commercially available, spare parts and supply chain. You just put it all together and it flies. You put another one, it flies again. So, oh, look, Mom, I'm a drone vendor right now, can get my first government contract. but, we still kinda, I believe a year away from, adding real smart features to the drone and, can we push hard in the direction. It, it should be different hardware, it should be different, different sets of the companion computers different sort of connectivity, lines. and, unfortunately Ukraine is not a good driver of innovations here by itself because, we still need, like hundreds of thousands of simple, FPV drones. We got used to, we are a little bit stuck in this point of history of the development. as was shown by Mr. Ford it's always not easy to bring the car instead of the horse because people always want just a better horse. But when there is a huge sense of urgency that's pushing you every single day, it's even harder'cause okay. That's a fancy new stuff. Amazing. Can we get another three hundreds of the previous brand new stuff? Because we got used to it.

Luka:

That's such an interesting point because on the one hand, the benefit of being in Ukraine obviously is the proximity to the end user and the ability to learn and iterate really, really quickly. But to your point, you're highlighting a propensity to prioritize short term benefit for, obviously the, the existential nature of the war, and the solutions that are needed that are needed today. And so it's almost that you need that Ukrainian footprint, but with a western innovation mindset and the western mindset to scaling the companies and building companies long term, do you think that this is, perhaps a, quote, a winning combination to bring innovation?

Serhii:

Absolutely. And one of the key things will, like I spend probably more than half of my time right now in Kiev and a lot of time on the business trip. So when I come back to Kiev, like you immediately like spend two days talking to all the same people, just right now in person, I, my planning horizon as a CEO falls down to work two weeks. And that's, it's, virtually unmanageable. So I have to get, out to visit Warsaw or to visit you guys or like go to Prague. And, I'm being a, now I am able to think like in years and think about the long term things just because the sense of urgency and just because things, that's a very, very good definition. The existential nature of the threat we cope with. that doesn't mean we don't have a good partners in Ukraine. We have, several armed forces, units and the brigades that are really good in, innovating and, pulling that to the real battlefield. but I really wish we have, we have much more, much, more of them.

Luka:

You mentioned that we're about, one or two years before we see some smart features being introduced into the drone space. What are those features?

Serhii:

I'd say it's scale. I'd say it's scale. So, because we have the very first deployments, we see the first deployments of the, of the autonomous, targeting of the autonomous ISR of the some complicated, things that, trying to describe what's going on there. But I, we still, keeping in mind the nature of the, of the business, keeping in mind the nature of the end user and the risks and the stakes. We have to be very, really, really careful here. And every single time when you failed with the very first adoption, it means that there is a word on the street that this technology doesn't work. And, this army, everybody know everyone. So kind, the word on the street works pretty well here. On the other hand, what we should have here is some different sort of, common operations even, let's put swarming aside, of group operations. So why the same direction? Why some formations? I do believe that, we will have a lot of improvements in UGVs because for ground vehicles, the navigation is still unsolved. And this is the key blocker for any kind of automated application or mass application.'Cause in contrary to flying drones, ground vehicles unmanned ground vehicles, cannot navigate terrain by themself. So they stuck in bushes they like, exposed themselves on the mines. they found the unmapped rivers and dive there. So all the, all the fancy stuff. So that's why we, we cannot automate this efficiently. And, as an army, we have hundreds of pilots who work with them.

Luka:

You mentioned earlier, operations Spiderweb.. I'm curious what are your takeaways from that.

Serhii:

I'd say my key takeaway here and my key outcome is the 99%. Okay, 95%. But nevertheless, the absolute majority of the complexity of the operation was not that drone by itself. That's logistics preparation, that's security service, and that was a brilliant operation and, making that happen in Russia especially, which is like by design, the very, very vertical integrated and, state is, just, that's, that's fantastic. But the technology by itself, the technology by itself is not something very, very unusual we can see here, in the set points, but the way, how it worked, and, I think the Secret Service is very professional and lucky at the same time.

Jim:

So it wasn't necessarily a spectacularly innovative drone capability that enabled the operation? The innovation was in other areas.

Serhii:

Yeah,

Jim:

How they got them in, how they

Serhii:

Abso Abso absolutely. Absolutely, because okay. And the LTE based drones, like, even our technology, we implemented that like a year ago. The terminal guidance that was supposedly used, I don, I don't want to touch the point, the terminal guidance on whatever else, because that's a very speculative topic. Again, it's been used for a year on the battlefield for different kinds of drones, but, that's not about the piece of technology. That the part, that's the state of the art of the operation in terms of the secret service operation. And I'd say we've seen that many times in the movies, correct? And there is some drone attack or some like trucks with the swarms of drones, et cetera, et cetera. the point that, first of all it happened and you know, the funniest part of the story was the one, the operator who was filming basically the truck, when the drones were flying out one by one, and he's just standing there filming the footage. And my first thing was, guys, don't you realize that there were like 55 kilos of explosive you in track in front of you, run, run out there? Because nobody knows, like, if you, if they see you have 20 drones, it mean two of them will never take off. And you can have some very unpleasant sequence for you. But I think that's a very good reminder or not reminder, probably. Europe lacks the wake up call, buzzword for the last year, but probably it's a wake up call for civilians because in contrary to arms, in contrary to rifles, in contrary to any kind of like pistols, whatever else, you have a drone, drone assault is virtually untrackable. So your operator is five kilometers away, sitting just in some, I don't know, from using the computer. So you, it's really hard to track him. that's different kind of threat.

Luka:

I agree just highlights an elevated level of, vulnerability of infrastructure that people have not really spent too much time thinking about, for better or worse, at least the broader population. Okay. What about, some of the technologies that you've seen over the last couple years that turn out to be tactical disappointments despite all the, excitement early on. You touched on terminal guidance, perhaps, we can put that in that column for now, but what other things would you include in that list?

Serhii:

Unfortunately, it could be different kinds of non-GNSS navigation and, many of them they were like from, stars, visual N Stars, like ra, whatever else, how to all, based navigation to all kinds of visual navigation, to all kinds of non GPS, whatever those receivers, because, Yeah, finally, many teams figure out why it took so many time and efforts to launch the GPS. It's, it's for reason and it's there for reason. And Baidu for the same, for the same story from Chinese. So it's far from being solved, We use it, we use it from several vendors, for instance, for our solution. But it's far from being solved completely. I'd say that, different kinds of automated systems and counter UAV systems failed. We've seen many of them from shooting some nets to kind of kinetic weapons, laser weapons, electro magnetic kind like microwave friers whatever else. Many promises, zero deliveries as for a moment. So we just start to scrape the final to scrape the ground. And it's done mostly by primes, not by small teams to have a complete automated, counter UAV system. One of the reasons because, and the core reasons is physics, that drones are really, really small with a small footprint EW footprints, and it takes you a lot of efforts and science to build the radar just to guide the counter UAV system. Other than that, other than that. Not so much. You know, I'd say, we still, as we still, uh, at mass use drones as, remote controlled grenades. So they are still remote controlled grenades with the camera.

Luka:

On counter UAS what are you most excited about in terms of effect, but also something that a startup is in a better position to develop as opposed to the prime? I.

Serhii:

That's a good point. Like as a, as the CEO of the startup who does, who does swarms we, we thought a little bit about the counter UAS systems and counter UAS for us is one of our, behaviors as well, obviously like swarm to counter swarms. but talking about the ideas because I've seen so many of them, I'd say, that any kind of, fancy next generation, things non kinetic, things kinda lasers kinda microwave, kinda whatever, like, sound waves, it isn't something worthwhile just by from the physics background. So, we still lack a small portable precision guidance system even for the, even for the rifle. So as soon as you can get a radar, sound, radar, whatever, like grid or connected network of detectors that you can shot drone for 20 meters away, it'll make a lot of difference.

Peter:

As you roll out this automated capability that SWARMER is working on, what are the implications of that as it gets employed at scale and what new challenges does it present? I mean for, and it goes all the way down to some of the core assumptions of how the combat environment is unfolding today. I mean, today we're dealing with relatively static front lines. We're dealing with a 20 to 25 kilometer gap between them that is covered by these drone systems. yet at the same time, we've just witnessed an attack that occurred, you know, deep in Russian territory, far, far behind those lines. And so if we intersect some of the different technological trends and how they're being employed operationally, as you roll out this level of coordination through automated software and connectivity, I mean, I could just imagine there are, a really large set of potential implications, and you're in the middle of this. You might have some really interesting insight as to what that's gonna look like.

Jim:

Could I ask maybe another question that precedes it, a little more elemental maybe, but important. What drove the need for what you're doing? What's the problem in the market right now in the battlefield that is compelling you to do what you're doing at Swarmer?

Serhii:

Oh, that's easy. Like page one of my pitch deck right now. But, thanks Jim. The point is, there is a single human pilot, at least one single human pilot behind every single drone for bigger drones. It's like three to five of them, literally. And it means when you try to deploy a million of drones, you need a million of pilots. And, if you're a manufacturer, want to sell million of drones. Army cannot purchase it because they don't have a million of pilots. And we solve this problem. We, change the way how it works. You, you're not an FPV pilot anymore, you're an operator and you work with as many drones as you have. So you know, asymmetrical advantage. Like another buzzword for us here is not send 200 drones. It's just send as many drones as you have, like everything from the stockpiles. And they will bring some value. Every single drone will bring some value. And you don't have to have thousands of pilots And this is a problem, and this is a problem we're solving. There are some pleasant consequences you don't have to have operators close to the front line. And basically, this is our like motto, send drone, keep humans end of the day, that's all about like robots versus robots. And when the robots stockpiles will be depleted, there will be a infantry, the foot on the ground.

Peter:

Okay. So if you are relieving a human operator bottleneck, which allows you to put a lot more drones into the air and to sustain that tempo, for a longer period of time, because you are not bottlenecked by human operators on sticks with goggles, then you could mount an operation, at some point in the combat theater on those lines that hasn't been seen before in terms of its level of intensity and its duration, which could be some of the implications that we're talking about here.

Serhii:

Yep. And I think the second one, here is the latency,'cause one of the key differences between, imagine you have 10 drones managed by 10 people and swarm by 10 drones managed by, by itself. Okay. And, the time for coordination and cooperation between human piloted drones is really, really awful. In contrary to human pilots of the aircraft, you don't see each other, you can't fly in formation, you have to look to the map to understand who does what. You have to talk to other people. You don't have visual contact. And, the human communication is, is really poor here is a real bottleneck. we figure out many from the data we have, figure out many cases when pilots try to like fly in swarm work in swarm. exciting, exciting insights On the one hand, on the other hand, never more than three guys, because, it's really, really tough to coordinate. And, let me try to, to tell you this story, how it started like in May, 2023. With computers, and this is exactly what I'm talking, like, we design system for computers, not for humans. there is virtually zero delay.. how did we discover the very first, real combat applications? We were sitting in the restaurant room with the, with the pilots, with the professional pilots, and I was saying, asking them guys, look, there were three or four professional pilots and two guys from the ground teams, ground teams who in charge of the ground deployment of the drones and the preparation and say, okay, imagine you execute some mission together. We spent 15 minutes just, speaking about, oh no, we don't work together. We're professional powers. We're like ACEs, et cetera, et cetera. Okay. You work together as a group and you have some complicated, like let's pull some example from your real application, from your real combat task, for instance. Destroying some, some I. Some buildings, with enemies, like, okay. And, you, there are four of you. You have four different drones or four different sizes and different kind of powers. Just was, try to make it a little bit, have, bring some fun here and imagine you can, take a break every single second. And as soon as you need to talk to each other, you just press pause and stand up and go take some cigarette, bring some coffee. You have two hours. It's like you play the game. We spend another 30 minutes just jumping into this direction and then. We figure it out, we started to figure out that there are many things they have never time to discuss, they have never time to negotiate. And okay, if there is such a complicated point here, I can ask you, because now I know that you're nearby, I can ask you to bring some high penetrating, explosive there, and then I will bring, another one. So, but it'll take them 10 minutes just to align on that, point that this is be, and this is the reason why they never do this in real life, but for computer it's like 15 milliseconds and as soon as it, the scale grows for computer, it's still the same 15 milliseconds for humans. It's, oh, it, no, no, no, no. Thank you. Thank you. That, and, this is, this is an interesting story because this kinda latency or, handicap or I don't know, or text for communication grows tremendously with the size of the group and it's even not about, I think, it's about both things, about the fear and about the difference of what drones brings to the, more than civilian security. Like there is a someone who has combined video camera and explosive where combined the capability and here is the same swarming brings you their instant cooperation at any scale. And hopefully it'll put us to the position when, it'll be a significant threat by itself that it'll prevent us from entering the actual warfare because, you can do things at scale with like 10 or a hundred x more efficient and more fast way than without, good global kind of computer based code.

Luka:

Before we start talking more about swarming, I think it might be good to, define some definitions and make sure that we're on the same page because, there are many people who have their own definition of what swarming is. You mentioned early on that some confuse operating multiple drones from a single controller as a type of swarming, which in fact it is not. And so in your textbook, what is swarming, what is not swarming?

Serhii:

Okay, so we have our definition as well. So, swarming is not automation. swarming is not, switching the one operator between many drones, or, swarming is through, collaboration and self-collaboration and self-communication by drones. So they work together to achieve the goals that humans set or another drone set probably in some future. But this is the point. So you don't task and you don't control a single asset. You just you operate the goals for the whole group and the timeline the sequences, whatever, like I imagine it's like I really, I really wish, and it was started not from just the defense on the company. We want to work on the cooperation of the normal robots in civilian world. So it's like I want to get my lawn I dunno, fix it here. I want to, start irrigation then, or just like do anything, do everything you need, to, to fix my backyard. And they manage by itself. So this is the way, how do you operate with a set of smart people that are able to coordinate together? And this is what swarming is for us, how we define it. Not about like, you do this, you do this, you do this, you do this, you have five minutes. That's not swarming, that's just, micromanagement.

Luka:

Right. Are you comfortable talking about certain conops that you are working towards or trying to enable?

Serhii:

I think we can touch a little bit, the very simple ones. so combined arms operation, the very simple combined arms operation called seek and hit approach when you have set of the drones, part of them are ISR drones, I mean intelligence, one, the rest are the strike ones. And, you set the area. Exactly. this is exactly the definition of how do you operate with the swarm. You set the area, you applied the weight of seek and hit operation and set the group of the drones like you have 10, 15 drones they take off fly there like manage all the stuff around, trajectories around like, you don't think about this. You just put, guys don't fly there. Why? Because they're a friendly, anti-air system that they just don't give a shit about, friendly or non-friendly forces. They hit everything that flies. or don't go there because there are enemies. But, other than that, it's up to you. And, as soon as you, as a human pilot has a video stream from one of the few, drones, you just, work with video stream and engage the targets. Do actual the pixel lock and then you don't care about which drone will hit the target. You just can say, okay, this is the vehicle and this is the infantry. And, if drone knows that it has anti infantry payload, it'll take a target. and vice versa. you don't think about whack. You just operate with the system as a, you have, you have five shots, for instance. and you have like three shots left so you can engage three more target. this is one very basic, very basic concept. Still has a lot of human in the loop, to give a sense of control and to keep, keep the safety as a P zero because for the 10 to 15 kilometers, there is a higher chance that friend or foe system, like, I mean, not, informational system, but the human and management system to detect is, can fail. And the second concept, again from the simple ones is the kill box. So, or kill zone is a NATO standard. So you define the area 35 kilometers away, 500 to 500 meters. kind of the point here is I as an officer is human officer is in charge. And this is my responsibility to say that there is only like no friendly forces, not civilians, where I take care. I take this responsibility. You go there, find all the vehicles, in that area, you know that there're not destroyed before and hit them and that's it. Five drones, 10 payloads go away completely autonomous like sending the squad of, special forces.

Jim:

I loved Luka's question. How do you define swarming? Because it is not what I would've thought. That's interesting. It's automated coordination is what you're, you're defining it as, as opposed to overwhelming automated coordination. But is there any automated coordination happening today as you define swarming?

Serhii:

I okay, probably somewhere in secret service or, intelligence agency. There was something, but

Jim:

Not in Ukraine. Not in, not in the

Serhii:

fortunately, for us, fortunately kind of for us as Ukrainians, I didn't see anything. And again, I, for the, for the last two years, my favorite joke at the pitch deck, like, I can write a book of such a size why nobody did swarming before. It's so easy. No, it's not easy. There is a huge set of technologies that are very immature and we still struggle with them. But no, we had a demo with one of our partners in, North America. I, I'm just trying to emit all the sensitive names here a month ago. And they were really, really excited because they say finally, this is officially the very first real swarm that flies, like mid air in North America. Everything we've seen before is just the automation.

Jim:

interesting, what's the level of difficulty of automated coordination of two versus a hundred drones on the battlefield?

Serhii:

It depends on how do we approach it. And this is where, why we are pretty unique. So we, we keep this mental model from day zero kind of, so that's why, our goal is to manage like hundreds of drones. We have a mathematical limitation right now, 690, but as soon as we face it, we can fix it somehow. Let's face it first. but, It depends on the approach with our solution it's, more on the communication network side, et cetera, et cetera. less than the mathematical one. and this is one of the main complexities here, but, as you talk, look, if you don't, if you don't touch swarm, if you work just with automation, this is where Chinese drone shown shine or Intel, drone shine, we have, six thousands of drones. You just, there is a special software that exists for like a decade and it works pretty well. You put a separate mission, like the, basically this is the root for every single pixel, and it's done worked pretty well.

Luka:

What is most difficult to develop, mature, execute on to reach swarming as a capability?

Serhii:

You know, the, I, I don't think that's the key, the real art of the story here, and I'm proud of, of my team and of my approach is to build it all and to make it work together based on the set of very, immature technologies. You work with, non GPS navigation, usually when you take something that like proven to be working on the combat or in like some kind of NATO country, usually. I don't know if I can pronounce this in your, but it sucks and it, it fails to deliver it. when you put it to the testing field, it works like five out of 10. And then when you, you manage to make it working, you still figure out that there are a lot of unpleasant side effects. The same with terminal guidance, the same with any kind of, target recognition, any kind of object detection, whatever. The so is set again of very, very, very poorly prepared technologies because they are there just for a year or two. There was nothing three years ago. And to put this all together and to find the way how the whole technological chain, is being built in the, in the way when every single next step compensates some issues with the previous step like we do with, for instance, with fire aid. When the team developed targeting stage, they knew that, the navigation system will probably miss the point for 300 meters. and they have to implement a set of special strategies to cope with that. And then it started to work together. And there are many, many little things like that when you, and that's complete autonomous system. Like, another dimension, we touch it as a beginner. So drones are still being built to be piloted. It means if something goes wrong. So there is a human pilot who can jump in, override the control and, fix the situation. When you have 25 drones with two kilos of payload above your head, no way you can just, you cannot jump in and, and fix all the 25 drones. It requires different approach to R&D requires different approach to testing, to the huge set of fallback strategies. Kind of what if things go wrong? What if like, I see and face this kind of malfunction and this gives a lot of more complexity.

Luka:

What is the most common pushback that you are getting from the engineering minded folks? Is it about how do you achieve, swarming in a, gnss denied environment? Is it about how do you achieve, inter swarm communications and collaboration in EW, type of environments? what's the most common pushback and which one is the most valid and the one that you're kind of, emphasizing.

Serhii:

The most common pushback is nothing works when they, trying to bring some new, for instance, we build integration for some scenario. And we are looking for, again, we are very focused on what we do, aka we don't try to reinvent everything. It means we use a lot of third party vendors to be fast. But, the keep push back is, oh, fancy new piece of technology. I bet it doesn't work. this is our experience for the last three years, but, the most valid one is probably, is the navigation because we work with many teams and we checked more than 16 teams to select the vendors we work with right now. And, as we're not building Anduril here, and we not don't have a budgets or a financial model for like 50 k, US dollars or euro per unit for the navigation, we should work with different kinds of the technologies, again, to be able to roll out at scale. And, we work with 16 or 26 who promised to deliver us best visual base navigation world. And it's, it's some kind of, street legend in our company. Or another one. Another one. Okay. Okay.

Jim:

What's the greatest value to the battlefield commander? Let's say they, they now have a Swarmer capability. What's the greatest advantage on the battlefield? is it that I can bring more operations with, well obvious, and obviously one. one is I only need one pilot. I don't need a lot of pilots. But if you were gonna talk to the battlefield commander, he or she now has your capabilities, describe to them what they're now able to do and at what cost that they weren't otherwise able to do in the past. I.

Serhii:

The key point, like the very, very instant feedback is as soon as I could put the pilot out of the front line. Okay, I'm done. I. So it, it's even, it's even, for swarming that's not the key, benefit, but one of the, one of the outcomes because you don't have to be a professional pilot anymore. And we build the whole system in this way. And, the second point is the reaction time. So, the ability to have several drones in the same place, like completely automated. So you don't think about how do you get them there? How many pilots do you need just to bring to that location, and then, have a very short latency time to kill or the kill chain latency, it brings interesting, that's one of the most important things for the army commanders, but this is the thing we, failed to quantify in any kind of doors because it saves lives. I would of lives and when I just say, guys, you can use five drones, like every single next drone you'll be able to use within every single next like 10 seconds in the destination area because we will bring them all together. They say a shut up and take my money. we quantified the efficiency of the drone, per strike in terms of the target versus, drone price ratio. But for this kind, we still struggle to measure it in terms of pen in, in, in, in value indoors because it's quite, quite hard.

Peter:

And to drive that point home. Help us understand how long does it take to train a drone pilot and, you know, what does that sort of progression look like for them? So that we understand how much of a change we're talking about.

Serhii:

I mean, when we introduce our system, that's interesting, because if it's a,

Peter:

I mean, even, even as the baseline, you know, without your system, how long does it take to train a pilot to be really effective?

Serhii:

So again, taking away any kind of our solution, let's talk about drones. So if you have a talented one, it takes you around two to four weeks to be a professional of FPV Pilot. That's a piloting part. Then you have to take, depending on the class and on on the manufacturer of the drone, is it kind of small FPV? So that's virtually zero, and it could be up to two to four weeks for the big ISR drones, you have to pass training for the, kind of check preparation. How do work with kind of, with boxing and unboxing? How do you work with many little sensitive things like antennas, et cetera, et cetera. Because most drones are still made of like many hanging pieces of plastic antennas and wire. You have to be, as a pilot, you have to be able to repair and to do the field, test, field fixes, for most of the equipment, if it's fixable in field, it's way better to do it right now than try to pack it, because. The exposure time and the ground that has field matters a lot. Every single minute increases the chances that you'll be hit by the Iskander or some other, rocket or FPV drone. So the numbers are like this from, I'd say from two to four weeks for the, for the training to be pilot by itself, depending on the class of the drone and on the, on the candidate. And the, the same amount of time for the, kind of more engineering training. How do you work with this specific model of the drone?

Peter:

But isn't there a lot of practice that has to go into flying these drones that takes more than a couple of weeks to become proficient at?

Serhii:

Yeah. Fortunately you can do it in simulator, so you don't need the actual drone. That's way easier.

Peter:

Okay.

Jim:

What's your greatest vulnerability? What's your greatest fear of a, you know, counter drone capability when you're developing coordinated automation like you are right now? And how is that vulnerability different than today's drone operations?

Serhii:

That's a good point because the difference between counter UAV for the swarm and the counter UAV for the one drone is basically the real time of the system and and the capacity of the counter UAV system. So that's more about the physical, features of the drone, the materials, kind of the dimensions, et cetera, et cetera. And the velocity then for the swarm itself, because of course there are like basic behaviors, like strike the same system from different, different sides at the same time, if you know the limitations of the rotation platform. But that's, I I, it's a shame to call it swarming and it's just automation. So that's a part of the embedded, the embedded, script. But that's a good point because, we are, we just, we just, literally, scratched the ground in terms of the real counter UAV systems. We just got the very first deployment of anti Shahed drones and they, far from being perfect, and a little bit expensive. So, we have virtually zero deployed, like just very, very several sets of the experimental deployments of the ground counter UAS systems for against FPV drones. We are not there yet, so, but we'll see. We started, that was interesting. We started a year ago to talk with Armed Forces Union saying, guys, we are trying to put some smart trajectories and some smart behaviors to harden the counter UAV approach, for the current deployments. Can you please share us some, good practices? how do you train your people to do, to, to counter uh, incoming UAS? Okay, we'll do this. So just fire everything you see right now. Fire everything you have from all the rifles, all the arms, and God bless you. Yeah. Okay. That's, it's tough to counter, but we will, we will try.

Luka:

High powered microwaves are often, described as a good counter UAS system against swarms. but at the same time, there are ways that one can shield electronics and another sensitive components. Do you see that as an effective weapon system against drones and swarms in particular?

Serhii:

I think that's probably the only effective system in terms of the, all the technologies we have. So laser but it's even easier for you to counter the laser just by pains, by the smart, smart configuration of the, of the surfaces. But I think, if you have the, phased array antenna, as emitter one and have a proper amount of power you can be fast enough to counter the swarm. the point always bear is, that's, that's a never ending never ending story. but I think in terms of the efficiency, it'll be one of the best solutions or, or a swarm of drones to countries form of drones just as easy.

Luka:

How do you expect the acceptance or adoption of the concept of swarming among militaries in the west to evolve in the coming years. Because, one can argue that militaries, yes, they are looking at the role of drones in, in Ukraine, but at the same time they are just struggling to figure out how do we institutionalize those drones? How do we change the doctrines? What is the impact on training? What is the impact on organizational structures? And they're yet on that first stage of embracing drones as a weapon system, let alone think about some more advanced conops like swarming. What are your thoughts on that and what do you see as catalysts to accelerate that sense of urgency?

Serhii:

So the first question was about what do I see in terms of adoption right now with the perception correct. that's interesting story. On the one hand, swarm, is a really buzzword of 2024 and a little bit of 2025. So I'm proud, to name the company, in the proper way. On the other hand, sounds like we are one of the very few companies who really understand what should work and what, what is just the fancy promises in terms of the, in terms of the, you know, they still, I. They or we, I don't know, like we are still figuring out as, as kinda a set of west countries, what's, what's going on there. Because for Ukrainians, this is the one, end of the scale kinda. We just need something right now, like in tens of thousands of units and just tell us how to use it we will use. In the West, that's different perception We have the air superiority we have artilleries we have everything. good luck trying to hit us with the swarm of drones we will just hit your factories and depots earlier. I think it'll be way better because as we work right now with several MoDs to, to get deployed, to get the first kind of conops. Really, approved ops concept, et cetera, et cetera. I think it's just too earlier right now to to talk about any kind of, specific matters.'cause everybody like French MoD, Uk, United States, Japanese, polish Lithuania. Everybody tries to figure out what the heck is going on what the heck should we do with the drones? And one of the interesting stories is the warfare nature is very different. Ukraine is very special kind of place because this is not something we see in Israel. This is not something we expect to see. I hope that never happened by, in, in Poland or in Estonia or in, Middle East, whatever. There's different, landscapes, different everything. so there is no, no good doctrine right now. There is no good set of strategies of applications unfortunately.

Luka:

When you say that those militaries are looking at what's happening in Ukraine and trying to, parse things out, what is it that they are having trouble with internalizing or observing? Do they question the relevance and applicability of this outside of Ukraine, or are they questioning this, disruption from such a low level of technology and weapon system? What is it that they're mostly quote confused about?

Serhii:

I think this is a good one. and I believe that the, or this is one of the reasons why we still have a lot of, a lot of, promises and the lot of, excitement around NATO Innovation Fund. I'm sorry for that. Around all the budgets for grants, for everything. we had the hard discussion in, way to 59 or two weeks ago about that it's way better to give money to Army Forces Unit and they will just get some experimental purchase. Let me try to explain. They, they try to put some, some money to the hands of the small startups because, what, what's happening in, in Ukraine, You can validate everything kinda right now. And, when you start to deliver something to the front line, you as a small, for instance, as a small startups from two students, you get connected to the actual, military guys. And you figure out the whole wonderful world of the consequences and all the surprises around you because oops, they re joint arms operation. You have to get intelligence you have to get confirmation, you have to get what, check everything that's you've never seen before. You cannot touch it in civilian life at all. There's different set of practices, compliances, et cetera, et cetera, et cetera. And, on the one given that, on the one hand, here in Europe, kind of, we try on the United States, we try to build some innovative, sandboxes. On the other hand, they have still miles away from the, from the actual armies and from the actual problems they're trying to solve. On the other hand, the army by itself, I have such a feeling that they look at Ukrainian kinda tens of thousands of drones. They are very, very low tech technology. And for instance, Germany army. Imagine I try to purchase 10 thousands of drones, for FPV drones. What I will do with them within the next year, because the batteries will die, I would not pass any compliance for any kind of hazardous materials, because they, that's different story. Zero training involved, no long term support. That's just kinda, there's different set of standards in terms of technology. so it's kinda a dead end. On the one hand, you cannot adopt and you don't plan to adopt what you see right now on the battlefield. On the other hand, when you try to innovate, but, there is no clear understanding okay, innovate what or what for.

Peter:

Let's talk about the startup system in Ukraine and where we see it evolving in the future role that we see it playing globally, right? There's a lot of talk about the innovation coming out of Ukraine, and yes, right now there are a lot of elements that make it. necessarily short term focused, but it's going to have a big impact. What do you see the shape of that being over the coming years? I.

Serhii:

I think we are following obviously the pretty normal, track of, developing any kind of, The pricing technology. But the same was with biotech. The same was with AI. The same was before with like dotcom even. so we will see consolidation, and in contrary to many other technological directions in computing or networking, you have to have a real hardware here. So it means consolidation is something that is inevitable because other than that, there is so not so many end users and not so many budget owners. and you should be delivered there mostly like as a startup, mostly, as a, as a piece of the hardware or as a part of someone's piece of the hardware. This is kinda, again, if you want to try to build a new company in 2025 for defense, target some component market, do something that's useful for many vendors because out of 2,500 world manufacturers, I don't think it's a good idea to be another one. so, and there are always first signs of movements and the big companies try to pull local teams and small teams. We like to purchase them to just invite them, to onboard them, to hire, to hire whatever else. We all here face the same story like it was in American early technological revolution era, when the bigger companies right now, they build a good and, successful hardware manufacturing business. They try to pull some small teams and startups to speed up the innovation, but that's still the same big, manufacturer with a very, very kinda, with a very, very proper culture. It's not about innovation. It's not about just doing things again and again, period. but I'm really curious how to take your look guys, because you're observing this from the well from the recipient standpoint, and you see how we changed it over the last two years.

Luka:

I would agree with the short-termism. That really, is evident when talking to a lot of teams. This sense of incrementalism is real. Also, a lot of, technical competence in the teams, and yet blind spots in terms of the ability to run a business, to grow a business, to scale a business, to, attract interest of investors. And also, the projects that, entrepreneurs are trying to build companies out of when in fact it is a, you know, it's, it's a piece of innovation that will be a part of a, a stack. and so those are some of the things that I would highlight.

Serhii:

What changed the last two years as, as you started to work with, as far as I remember, or year and a half, as you start to touch, the Ukrainian market, like any changes for negative direction or for positive one,

Luka:

I think people started to ask for more money.

Serhii:

Is it a positive one or a negative one?

Luka:

It depends. It depends. I think, you know, people have gotten, better in, describing the value proposition. I think they have learned, the, VC ecosystem and that aspect of the dynamics is a lot more, understood than earlier. And I think thanks to, some of the programs that exist in Ukraine, accelerators or otherwise that are, coaching entrepreneurs. I think there has been a, maturity on that front.

Serhii:

Okay, that's great. So, kind of phrasing it, people, we here at the, this side of the story, startups, start to figure out better that, in Ukraine that, VCs have your own metrics, you have your own goals, and, kinda that's, that's not a kind of grant of free money. that's a way of doing business in a proper way.

Luka:

Mm-hmm.

Serhii:

Cool.

Luka:

that's right. And, I think, even more understanding of the VC model is necessary for founders to be able to then understand the investor's perspective and the criteria that, folks are like us are evaluating and assessing. Not everything is a venture backable business. Certainly most companies that we have talked in Ukraine are not, despite the interesting work that they're doing. There are other sources of capital that are probably better suited for those projects.

Serhii:

Perfect. So we are, we are learning slowly but surely. Oh, that was, that was a great setting.

Peter:

It's, it's like trying to find long-term differentiation amongst these companies combined with applicability to markets outside and beyond Ukraine,

Serhii:

Mm-hmm. That makes,

Peter:

markets, three years, five years from now. Where do these technology roadmaps fit in to that future picture? I, I think that's been a, a, a very common question that we've come up against.

Serhii:

yeah, makes perfect sense.

Peter:

I.

Luka:

Any parting thoughts? Serhii for the audience?

Serhii:

Not really. I think we touched almost everything. Almost everything. So, I really hope that, in a year we'll have the same conversation in a very different set, in the peace time and talking about how should we apply the outcomes of the, highly developed the defense sector in Ukraine to the civilian world now,

Luka:

Likewise,

Serhii:

Yeah.

Luka:

likewise. Yeah. Fascinating conversation. Thank you very much for finding time

Serhii:

thank you very much for hosting me. Was was a pleasure to talk to you guys.