--:-- Host (Erin Price-Wright): So I'm really excited because we have someone from the business, from the product, and from the tech side here, as well as air, land, and sea. Maybe to get started, I would love to hear how far along in the journey to autonomy you feel like we are in your industry. Maybe framing the levels of autonomy: where are we really all the way there, and where do we still have some way to go?
--:-- Macario Namie: So Skydio was started on a foundation of autonomy. We did not invent the quadcopter; those existed prior to the founding of the business. I think we've all experienced or seen them, even just walking around Best Buy in toy form, where you have some sort of radio-controlled mechanism to tell the drone where to fly.
--:-- The challenge with drones is that it always required a person to be there. The more you go up in the use cases into enterprise or government, the more valuable the flying becomes. If you're going to fly around a substation that belongs to an energy utility and you crash that drone, you could take out power to an entire neighborhood. That's not something cute that your kids did on Christmas afternoon. That's something that would cause you to lose your job because it creates real problems for the community.
--:-- So the necessity for the person is not just to be there, but they have to be an expert pilot. They have to be great at flying in really challenging environments. That was the premise of Skydio: can we build the skills of an expert pilot into the drone itself so that any one of us here in this room can pick up the drone and be reasonably proficient, democratizing access?
--:-- We started day one building in autonomy. The first iteration was the ability to follow somebody. Since then, we've upped our game quite a bit, not only beyond obstacle avoidance, but to be able to not have a person present at all. Going back to that substation example, there are now drones here in the Bay Area that are flown from somebody 20, 30, 50 miles away. Those drones complete automated inspection missions. So we're getting there. The regulatory environment doesn't allow us to completely eliminate the person, but we're certainly making quite a bit of progress.
--:-- Vijay Patnaik: In the automotive space, for passenger cars at the lower levels of automation, what we'd think of as a hands-on system where you're still paying attention to the vehicle, your hands are on the steering wheel, but the vehicle can do things like automatic emergency braking or cruise control—those systems are available on any car you go and buy today. In Europe, some of those are mandatory from a regulation standpoint. I think US regulations are slightly behind in mandating some of those systems, but generally they're available on vehicles.
--:-- The next focus area for a lot of OEMs is on what's called level two plus systems—these hands-off systems. As a user, you don't need to keep your hands on the steering wheel, but you still need to pay attention and keep your eyes on the road. One level beyond that is eyes-off as well, so you can read a book or watch a movie. Those systems aren't yet deployed; they're in R&D phase right now with just very early deployments.
--:-- On the level four side, in San Francisco, we've all seen Waymo's robotaxis. That's the main deployment on the car side right now. There are also some deployments in China on the level four side.
--:-- Taking a different industry, mining has had some form of trucks without any drivers in them since 2007, 2008. It's not the same level of autonomy as Waymo in the sense that those trucks are following predefined paths and get a lot of support from the infrastructure in the mine. But there's a clear business case and ROI for that product, so the mining industry has been investing in it. The technology itself is far behind what's available in automotive today, so there's a big focus on upgrading that technology and serving more use cases.
--:-- Peter Bowman-Davis: Maritime takes a lot of inspiration from the self-driving car community. In maritime, you have a two-dimensional plane that you're moving around, very similar to self-driving cars. You just have lateral planning. It's maybe a little more complex because you have the longitudinal axis as well, and you have much less things in your environment to ground you.
--:-- When you're thinking about an autonomy model, the world doesn't usually just nicely translate around you. Think about driving a car in a nice city skyscraper block. As you take a left, the world nicely transforms around you. But imagine you're in the middle of the ocean and you take a left. The ocean looks the exact same. So you actually get a lot less information per frame when you're making vision models or an autonomy stack.
--:-- The other thing I want to draw is the delineation between a perception stack and an autonomy stack. For a lot of these use cases, for us at least, perception has mostly been solved. Object detection, object avoidance—these are very simple tasks that can mostly be done with CNNs, which are 20 or 30 years old. Autonomy is trickier because you actually have to take elements from the perception stack and make them actionable. You have to do a lot of long-term planning, take in that data, and make decisions based on it. That's something that's only been enabled in the last five years, maybe even two years.
--:-- I'm super excited about a lot of the work going into reinforcement learning and long-term planning because it's super important in the maritime domain. In maritime, you're not often near the shore, and in the defense application, you're not often in communications with back home. So you need to make a lot of independent decisions. These are the things we're thinking about in the maritime autonomy space.
--:-- Host (Erin Price-Wright): Double-clicking on some of those technical breakthroughs: where do you think the latest developments in AI have impacted autonomy? To what extent have they impacted autonomy, and how much change are we seeing in the industry today versus five or ten years ago, based on the current state of the art—things like video language models or other new transformer-based architectures that might not have existed when Waymo got started or when Skydio got started?
--:-- Vijay Patnaik: I think pretty significant impacts, actually. In terms of the latest developments, foundation models do have an impact on the architecture of these autonomy systems on ground-based systems. You think of foundation models as replacing a number of more task-specific models that existed. We see more and more companies making the shift towards using or at least researching how foundation models can be used.
--:-- The frontier of research right now is end-to-end. Even when I joined Waymo back in 2016, there was a lot of hype around end-to-end driving, but it didn't really materialize. The technology wasn't ready. Now it's almost a second version of that happening, where the research has progressed a little bit further. I do think the architectures are going to evolve significantly, even though they're not yet ready for production.
--:-- What that also means is that the tools and infrastructure needed to support that is evolving pretty quickly. We are developing new kinds of simulators needed to support those advanced architectures. Those simulators themselves require many of these generative techniques or neural rendering techniques that have shown good promise in research.
--:-- Third, I would say just using some of the generative AI to simplify workflows. As an example, if you're an autonomy engineer, you spend a lot of time working with simulation and scenarios, but you don't want to write a lot of complicated scenarios. We can generate them for you programmatically in a much more confident manner today than what was possible two or three years ago. So at different layers—whether it's the autonomy stack or the tools being used to develop that autonomy stack—we see pretty fundamental impacts. That's why we're investing pretty aggressively in utilizing these technologies.
--:-- Peter Bowman-Davis: Just double-clicking on the idea of simulation for robotics or autonomy: I think this is a very non-obvious point to people who haven't worked in this industry for the last two years. We've started to use video generative models as drop-in replacements for simulators like Unity, ROS, or Unreal Engine.
--:-- The reason you need a simulator is because robots are really expensive, and you don't want to take them out to the field to have them break if you mess up the algorithm. You also need many environments where you need to train on policy, meaning you need to run it in real time to make sure the thing works. In recent years, they've started to use video generative models conditioned on actions or sensor data, and you can actually create a really incredible reconstruction of the world just using these video generative models. This is not hype Gen AI stuff. This is used by Tesla. This is used by comma.ai. It's pretty exciting stuff. I think some people are being very quiet about it, but there are some really awesome public releases.
--:-- Macario Namie: I will say the rise of NVIDIA has been pretty instrumental for our business. To be able to put that kind of compute power in a four-and-a-half-pound form factor that can be sold for basically $11,000 or $12,000, and we'd still make some margin—that's a very hard thing to do.
--:-- Host (Erin Price-Wright): I'm curious, how do you think about the cost aspect when using the high-performance chips that NVIDIA comes out with? How does the ROI for the business change when you're putting the latest GPUs or compute onto the drones?
--:-- Macario Namie: It's so foundational to our business. We have to have high compute on the drones themselves. In our latest release, which we announced about a year ago, we actually added excess capacity—basically trying to future-proof the hardware so we can do more over time. We haven't fully used it. We expect to be able to run additional models, including models that our customers build on the drone itself, because they can do much deeper analysis on the things that matter to them.
--:-- Host (Erin Price-Wright): What's an example of a custom model that a customer might want to build?
--:-- Macario Namie: Object detection is a really simple one. But in our world, we see everything. It's not just about people. It's like, "I want to be able to identify that particular transformer." By the way, PG&E's transformers are different than Southern California Edison's transformers, which are different than Baltimore Gas and Electric's transformers. So they know their transformers. Once we identify the transformer, we need to be able to determine whether the thermal signature is saying there's a problem or not. That is so specific. We're not the best ones to build that. So that's an example of how customers can build their own stuff and run it so it can take immediate action on the drone itself. That's why we put in that excess capacity—to have that kind of extensibility in the platform.
--:-- Host (Erin Price-Wright): Very cool. That brings me to another question. For Saronic and Skydio, both companies are essentially building the full vertical stack—not just the algorithms around autonomy and perception, but the full product that goes into the world. Whereas Applied, you're much more of a software provider working with companies who are building the hardware. I'd be interested to hear some of the trade-offs. What do you gain by owning the full end-to-end system? And what's harder as a result of having to manufacture and deliver the final product?
--:-- Macario Namie: We think the ultimate value of a vertically integrated stack is reliability. We're very committed to it. We started with it. Our primary competition in the world is DJI, the largest manufacturer of drones. They are the complete opposite. They're happy to build the hardware, and there's a whole plethora of other companies to build their ecosystem around DJI hardware. We see that those seams start to fracture when you get more and more complicated scenarios. You add in autonomy—what happens in the corner cases? What happens in the failure modes? How do things react? That's where we can completely control that, test for it, and build reliability around it.
--:-- The downside is that we might go slightly slower because we have finite resources. We have to choose where our investments are. Another company may focus entirely on building some capability on top of another person's hardware. For us, we're okay moving a little slower to deliver higher quality.
--:-- Peter Bowman-Davis: We're kind of a weird middle ground because we own the full stack. We also have three types of boats—small, medium, and large. These boats are very different in terms of the amount of cameras they have and the actuation techniques you use. That being said, we still own all the hardware, especially all the compute. The compute is mostly homogenous between them, but your actual effector systems or your perception stack is going to be differentiated. It's also differentiated between boats sometimes because we want to test things out, add another camera, or something.
--:-- I would say it's actually a powerful thing to be a little bit agnostic on the hardware and the software because it pushes you to create really solid abstractions for your perception stack or your autonomy stack. It pushes you to be a better engineer. This is at least what I've experienced in my engineering team. We really like to think about how this is going to work on N plus one. That's been a fun challenge to work on, but I do agree that it does slow things down sometimes.
--:-- Vijay Patnaik: I completely agree with that point of abstraction. Applied started as a tools provider. We would provide simulation and data products. Eventually, our customers said, "Tools are great, but we don't have the internal capability to build that autonomy stack using the tools. Can you help us with that?" That's how Applied actually started. We now provide an off-road autonomy stack and a trucking stack.
--:-- To the point Peter was making, now more than ever in autonomy, these abstractions are possible. We have a stack that is not built for a single customer. It's meant to be reusable across customers, and they can build their eventual application on top of it. If you only want to use the tools, that's great. But if you want an actual autonomy stack—either as your primary stack or to kickstart your efforts internally—there are providers like Applied or others that you can use with the right abstraction to significantly accelerate your program.
--:-- Host (Erin Price-Wright): How involved is that translation process to different form factors, vehicles, and customer types? I'm curious to hear how hard it is to get autonomy that works for one type of vehicle to work in a totally different setup.
--:-- Vijay Patnaik: I would love to tell you it's seamless and just happens by itself. But there's the engineering aspect and the organizational aspect. Engineering-wise, to a certain extent, you can engineer the best APIs and abstractions. There's still customization you would need to do, especially if it's the first time being deployed to different platforms.
--:-- The organizational challenge is even more interesting. Some of the companies we're working with are going through a big shift from being hardware companies—because vehicles were primarily hardware driven—to becoming software companies. They're reinventing themselves, hiring software engineers, hiring engineering leaders from Silicon Valley. You actually need a very close partnership to make this work, especially with various vehicle form factors. So it's not just an engineering problem. It's also how you build up internal capabilities for customers, often working with them on training their internal team, to work together on transitioning the stack to many different platforms.
--:-- Host (Erin Price-Wright): Makes sense. I spent most of my career before investing at Palantir, so I'm very comfortable with that type of model. What was your experience there in terms of how to help companies make that shift to autonomy, software, and data? It wasn't strictly autonomy, but I think there are a lot of similarities.
--:-- Vijay Patnaik: The biggest thing was how you embed with the customer to drive cultural change. That's the forward-deployed engineer model really invented at Palantir, where I would be flying out to Azerbaijan, Oman, Trinidad, or really far-flung places to help their engineers on the ground in the field figure out how to use the software.
--:-- Host (Erin Price-Wright): Maybe switching gears a little bit, I'd love to talk about the economics of autonomous systems and how that relates to regulation. We've heard and read in the news about Waymo and the economics of a Waymo ride. It's really exciting to see it start to actually make sense, so you're not just pouring money into a system for the sake of R&D. Every single Waymo ride is, I think, profitable if you don't count the overall cost of the vehicle, which is a big cost to discount. I'd love to hear, starting with Skydio, how you think about the economics of the drone industry, where you've seen real success, where it's been slower, and how regulation has played a role.
--:-- Macario Namie: We serve multiple different industries, and their ROI calculations are slightly different. With energy utilities, it's about looking at the people and labor it takes to go inspect equipment—physical substations, transmission lines, or distribution lines. Anytime you can save the cost of sending a truck on site, putting a person in a bucket, getting them up in the air—which can cost $2,000 per deployment—and instead put a drone up and get the exact same information in 60 seconds, there's a pretty obvious ROI.
--:-- We also serve public safety, like first responders and police departments. You can't really put a value on human life, but you can put a value on the insurance payouts that come with officer-involved shootings. It can easily go to one or two million dollars every time there's some kind of use of force. Being able to avoid that, to cut that down in half, is really key.
--:-- Ultimately, we're in the business of providing information for people to make better decisions. In a high-stakes scenario, better information about where the threat is can really make a big difference in whether you end in a peaceful way or a tragic way. Certainly there are elements of economics, but a lot of it is just keeping people safe—keeping officers safe, keeping the community safe across the board.
--:-- Host (Erin Price-Wright): And how about the regulatory piece? Whether an operator has a line of sight on a drone—take a particular use case like police officers. How do they think about fleet expansion as those regulations change?
--:-- Macario Namie: One of the key areas of investment right now is a concept called "drone as first responder." You pre-position docking stations on the rooftops of cities. This is happening in New York. It's being tested in other cities. When a 911 call comes in, the drone automatically deploys and goes, effectively being the first responder—the first person on site—providing eyes back to a remote pilot.
--:-- That does two things. One, the pilot is sitting in a nice air-conditioned office and can immediately relay information back to responding officers so they can respond more appropriately. They can control many drones at the same time, so it becomes a one-to-many versus a one-to-one argument. A lot of calls can be cleared. These 911 calls for service can be cleared by the drone saying, "There's no real issue here. You don't have to go on site." That's tremendous economic savings.
--:-- For drone as first responder to happen, you have to have a regulatory environment that allows for a lack of visual observer on the rooftop. A 20-second primer on FAA regulations: if you're putting anything in the air, it's regulated. The FAA cares about airspace, mostly about not hitting anything—don't hit an airplane or a helicopter. They require you to stay below 400 feet and require a person to be visually looking at that drone at all times. Those are the most basic rules.
--:-- Now you have a scenario where you might go further up. You don't have the available staff to put a person on the rooftop to look out. And if you're in New York City, the top 100 buildings are over 600 feet. How are you going to keep everything below 400 feet? Being able to work with the FAA to create waivers to this traditional regulatory environment—allowing for beyond visual line of sight without a visual observer, and to stay within 50 feet of a structure so you can go up and over a building as long as you're within 50 feet—that was a waiver that came out two or three weeks ago in New York City. It's a fantastic way of having the FAA work with these agencies to do what's right for the community while still keeping the air safe.
--:-- Host (Erin Price-Wright): Exciting. Congrats. I'd love to hear from Applied Intuition about the economic case. Where is this still a slightly money-losing R&D exercise versus where are you actually starting to see the economics make sense? And where, if at all, does regulation play a role?
--:-- Vijay Patnaik: On passenger cars, they've had driver assistance systems for a long time. As these systems become more capable, OEMs have an ability to charge more for them. That's why the FSD system from Tesla, when it came out, was over $10,000.
--:-- Host (Erin Price-Wright): As an avid Tesla full self-driving user...
--:-- Vijay Patnaik: There you go. I get a test.
--:-- Host (Erin Price-Wright): Yeah.
--:-- Vijay Patnaik: So there's a certain willingness to pay if the performance of the system is up there. Tesla has discounted that, so I think the industry is in a price-finding stage right now. Obviously, the best thing would be higher performance while keeping costs the same. But as the cost of compute goes up, more sensors go into the vehicle, more software is deployed onto it, the BOM cost for the OEM goes up. So I think there will be some price increases.
--:-- OEMs already operate on somewhat thin margins compared to the margins we're used to in Silicon Valley. They're not in a position to keep losing money on each of these systems. I think the market will eventually find a price that makes it profitable for the OEM to ship these systems.
--:-- On areas like trucks and construction mining, they have a different unit economics framework. If you think about trucks, for example—and I'll tie in regulation as well—we can think of unit economics as reducing current costs. But a big part of unit economics is also whether you can increase the revenue per truck or not.
--:-- The first argument is that there's a shortage. We're losing supply chain flexibility. We do some work in Japan providing our autonomous truck technology. In Japan, the government is actually pushing the commercial vehicle sector to invest more in autonomy because they're seeing what they call the "2024 problem," where drivers are aging out, overworked, and have health issues. So the shortage problem is real. But even if you look beyond reducing labor cost and insurance cost...
--:-- Host (Erin Price-Wright): If you have a network of autonomous trucks, you can actually optimize the entire logistics network, such that you're generating more revenue per truck, which makes unit economics better. Where regulation plays a part in that is twofold. One is in a place like the US, each state can actually regulate how these driverless trucks are deployed. And so you can have this weird mix where the states that your truck is traveling through have slightly different regulations and federal government has different regulations. So you do need some consistency. If you want to have a truck go from LA to Atlanta, which is a pretty heavy freight route, then California has to have consistent laws with Arizona and Texas that allows that. And I think it's not fully there yet, but getting there.
--:-- The other side of it, there are regulations like hours of service that determine how long a truck and a truck driver can drive. That's for their safety from fatigue, et cetera, and for road safety. So the government has to be willing to be flexible on those such that those trucks can then operate 24 hours, such that you can then reconfigure logistics networks and then make the supply chain more efficient and make the revenue and the unit economics work. So I think those are some of the factors in trucking.
--:-- And mining, I think, is a different set of factors. The ROI is somewhat already clear to the industry. That's why they've been investing in autonomy for a long time. The trucks we're talking about in mining on which these autonomy systems are being deployed, this is like a $5 million machine. So it's not a big deal if you put $100,000 LiDAR on it. Of course, it adds $100,000 and that's meaningful. But relative to a $5 million machine, that's not the main point. The main point is if the system stops and it stops the mining operations, for every minute you stop the mine, you're losing tens and hundreds of thousands of dollars. So the unit economics game there is at a mine level, not necessarily at a truck level.
--:-- Peter Bowman-Davis: I also imagine replacement cycles are different. It's not as easy to buy a new $5 million bulldozer as it is to buy a new $100,000 car.
--:-- Vijay Patnaik: That's right. But the whole ROI calculation becomes a slightly different calculation, which is how many trucks do you deploy in a mine? How do they impact the productivity, which increases the mine revenue? And how do they impact the downtime? So we see slightly different applications, but I think generally in mining, as I said, OEMs and the mines we talk to see the business case there is just improving the technology such that it can reliably work and not bring downtime to the mines.
--:-- Host (Erin Price-Wright): So I think related to some of the questions around regulation, I'd love to talk a little bit about geopolitics and some of the regional or global differences in our approach to autonomy. I don't know if this is a true story, but I did hear the story about DJI drones being used early on in Ukraine. And China had given Russians a backdoor to be able to determine the positions of Ukrainians on the ground using DJI drones, which was a big impetus for a lot of the drone activity that we're seeing in Ukraine coming from the U.S. So there's a lot going on in this geopolitical landscape. I'd love to hear how, for all three of you actually, given Saronic as a defense contractor, how the role of geopolitics and how you think about autonomy has come up both as a challenge and also a motivating factor.
--:-- Peter Bowman-Davis: Saronic was basically founded on this idea that China outnumbers our shipbuilding capacity about 200 to 1. So the question is, what can we use as a sort of unfair advantage to leapfrog that rather than just compete on the basis of cost per ship? How can we make autonomous distributed systems deliver an overmatched result over China's shipbuilding capacity? This has obviously played a major role in motivating Saronic as a company. But even beyond that, I think we've seen a lot of unmanned surface vehicles, which are autonomous boats, be used in the Middle East, as well as with Ukraine and Russia. The heating up of global proxy wars, not just great power conflict, has also led to a lot of expansion of use of autonomous systems, as we see in Ukraine, as we see off the coast of Yemen and other countries. But yeah, that's the TLDR.
--:-- Host (Erin Price-Wright): And then given the DJI angle here, would love to hear how Skydio thinks about this.
--:-- Macario Namie: Geopolitics is a very important topic in our world. You mentioned Ukraine. We've been in Ukraine 30 times. I was there this past spring. They use DJIs in a very disposable fashion, because it's the cheapest they can get and they hack it. DJI is effectively a hostile vendor to them. So they have to hack it quite extensively so that the drones can actually fly there and not be detected by the Russians. There are certain techniques they have to do that.
--:-- One of the challenges that the U.S. looks at is the threat of being wholly dependent on Chinese-based technology for industries that they would consider to be critical to national security. What Ukraine has shown to the world is how critical drones can be in any combat. And if all of our drones and all of our drone parts are being held by Chinese manufacturers, then we are wholly dependent on them for such a critical technology. That is exactly what the U.S. is trying to avoid. There are laws being discussed within Congress now to start weaning the United States off of Chinese manufactured drones. It won't be an immediate ban. It won't happen overnight, but they can start lessening their dependence and start encouraging U.S. innovators to step into the fold, including ourselves. We're not asking for subsidies, but we do benefit from restrictions that force some of these organizations like the federal government or state and local governments to choose Western products instead of Chinese manufactured products.
--:-- One small example of the kind of power that some of these manufacturers can have: Autel, which is a smaller version of DJI based in the exact same city in China, decided one day that they would update their geofence, basically where you're allowed to fly and not allowed to fly, and they turned off all access in Taiwan. Immediately every Autel drone was just bricked in Taiwan. If you had an example where all of our critical industries are using a Chinese manufactured drone and one day China says, "I'm just going to turn them all off," they could immediately brick them. That's really the extent of the problem. It's quite severe.
--:-- Vijay Patnaik: Yeah, I think China is a conversation with every CTO or CEO of any major OEM across industry that visits our office or we talk to. These are weekly conversations. To take an example from automotive and maybe government, because Applied also does a bunch of work with the government. In automotive, China is basically redefining the industry. There's no lighter way to put it.
--:-- The Beijing Auto Show was earlier this year. I happened to be there in person for it. All of the attention was on the local Chinese OEMs. You could literally go to the booths of the international companies and consumers didn't really have much interest in them except for a couple of brands. That's because the product innovation in the China ecosystem is second to none right now. The cars you can buy there, not just drive in a prototype, are vastly superior consumer experiences compared to what you can get in the US or Europe.
--:-- I was talking to Steph earlier about some of the experiences there. You walk up to a car and you say, "Please un-park yourself." It un-parks itself, opens the door for you. You sit in the car. The car assistant can process commands from four people talking simultaneously. It's a delightful experience to sit in those. These are production cars. And it's not just about the product today. It's actually scary how fast the pace of innovation is. Every six months, they're innovating on that product. And the cost at which they're able to do it is very hard for the global industry to match.
--:-- At one point, you might think, well, there's enough geopolitics that maybe the global economy is somewhat insulated from that. We're seeing that in the US where there's 100% tariffs on Chinese vehicles. At the same time, you look at what's happening in the market in China, where these OEMs are facing a flattening domestic demand, they have oversupply. So even though China is already the largest automotive market and the largest exporter, they all have ambitions of going global. Some of these OEMs have cost points that are so low, like a BYD, that despite tariffs in Europe and tariffs in the US, they might actually be somewhat competitive. That's why every automotive executive is worried about China and thinking about their strategy, not just in China, but their global strategy, given what we are seeing in China.
--:-- I think on the government side, it's similar. The DoD used to be the driver of innovation. There are a lot of ties between Silicon Valley history and DARPA and how autonomy came about and how a number of other technologies, like even the internet, came about. Now we're at a point where the Pentagon recognizes that it's been slow in moving towards software and moving towards autonomy, and definitely slower than China, both in terms of processes and how you do procurement. That is a big barrier today.
--:-- We do this conference every year in DC, focused on national security. The entire conversation is about how do we move faster and bring these technologies that are being deployed in the commercial world to national security much faster. And how do we evolve our procurement processes to be able to do that? So I think across industries, we see China being a major part of conversation, a major driver of strategy for companies.
--:-- Host (Erin Price-Wright): I was going to end with the last question of what keeps you motivated, but I feel like that's very motivating. So with that, I think we have a few minutes left. I wanted to just open it up if anyone in the room had any questions for our wonderful guests.
--:-- [Audience questions are condensed for the podcast.]
--:-- Host (Erin Price-Wright): All right, since this was a live recording and some of the questions were pretty long, we've condensed them for your ears, and I'm going to be punching in with a voiceover. All right, first question. What are your experiences dealing with security guidelines and regulatory bodies?
--:-- Peter Bowman-Davis: The one thing that I would say is it's a little bit unintuitive sometimes. Designing autonomous systems, as long as they're not true kinetic effectors, sometimes it's actually lower risk. The example that I like to give a lot is imagine designing a drone versus a helicopter. A helicopter, you've got to be pretty sure that that thing will not fall out of the sky. A drone falls out of the sky? Okay, that's fine. We'll iterate. We'll keep going. Sometimes it actually helps me sleep a little bit better at night thinking, okay, at the end of the day, these systems are directing themselves. If the boat sinks, it's not the end of the world.
--:-- In terms of the actual reality, the fact that we're designing for the DoD though, yes, these systems have to be absolutely secure, mission critical, especially in terms of reliability. To this extent, Saronic as an engineering org is built up around this. The way we do our systems engineering, the way we do our tests and evaluation pipelines, is a lot more rigorous than a lot of other companies. But at the end of the day, you can sleep a little bit nicer just knowing that they're unmanned systems.
--:-- Macario Namie: From a product design standpoint, in terms of safety, a lot goes into failure mode. So what happens if the battery gets too low? No matter what inputs the operator gives, the drone is going to come back and it's going to follow the path that it had. It can navigate back from obstacles no matter where it's at to get back to its original location. It can also land directly down. There's a programming of safe landing zone. Those are all things to try and keep people safe.
--:-- In terms of more security, especially with the Department of Defense, we sell a variant of our products that are effectively offline. "Offline" is a bit of a misnomer. They're online, but completely within the private network of the DoD. We have no access to it. It's not cloud. Those are instances where if the customer really needs that level of security, they have it. It means more engineering work for us because we have to carry two variants forward, but it's necessary in order to serve the federal government.
--:-- Vijay Patnaik: Yeah, generally in the automotive realm and Waymo cars, et cetera, what I would say is that what's been done in the industry in the past, all of the systems engineering practices that have been used to build aircraft like commercial airliners, are necessary and are used, but not sufficient. In the same way, some of the standards that have been put out, like ISO 26262, have been used in the industry for a long time. These are all inputs, but none of these individually or even cumulatively are sufficient.
--:-- If you look at some of the publications from Waymo and some of the work we do with our customers, we actually consistently tell our customers we need to do way more than what any regulatory body is asking for or what aerospace engineering has done in order to prove the safety of these systems. Validating that these systems are safe to be deployed, whether that's a passenger car, a truck, mining, agriculture, or defense application, is actually one of the problems our customers struggle with the most because there is no standard blueprint for it. It's a combination of an engineering problem, plus a data science problem, plus a regulation problem, plus winning consumer trust problem, because eventually you have to convince yourself and the stakeholders that this is safe enough. It's not a science problem, but in many ways, the blueprint is yet to be clearly written for this. I just don't think the regulations and standards are enough. The same applies for cybersecurity. I think they're very short of what actually needs to be put into these systems.
--:-- Host (Erin Price-Wright): What best practices do you use to reduce stress in your technical debt?
--:-- Peter Bowman-Davis: If you ask any Saronic software engineer what the three magic words are, they'll tell you this. The design principles that we follow are simple, correct, fast, in that order. The idea is that, first of all, you want to build a very, very simple system that you understand from the electron level. We do not integrate any technology that we don't understand from the electron level all the way up to the software level. The TLDR is that we build a ton of stuff that doesn't work at first, but that we understand super, super well. The systems design philosophy pervades basically everything in our software spec. So yeah, simple, correct, fast is the way we do it at Saronic.
--:-- Macario Namie: This has totally worked for my opinion. I've never heard of technical debt. What's that?
--:-- Vijay Patnaik: I think technical debt is inevitable in fast changing industries like the ones we play in. We were previously talking about how the underlying technology in these systems is changing, which means we have to rethink the products we are providing to the market quite a bit. You keep the current system stable and then you deploy a new system magically over time. But we've all been doing engineering for a long time to know that it's never as clear cut as that.
--:-- I think we focus a lot on working with our customers on joint roadmaps where we literally go to customers and say, "Hey, we're going to actually upgrade the architecture of this product because now we can make use of generative AI or any other technology in order to improve these products." Surprisingly, more often than not, if you have a reasonable roadmap, the customers are actually pretty reasonable. It does require you to have a trusted relationship where it's not just "here's some money and you're a vendor." If you're in commoditized markets where you're just purchasing based on certain features and who's the cheapest to procure, it's hard to have that kind of relationship. But for the industries we're in, where the software we are providing is cutting edge, and they know this is cutting edge software, it's a very collaborative relationship with the customer. We're often building new products with them. That allows us to have thoughtful roadmaps.
--:-- Macario Namie: Man, I've been in tech 25 years. There's always technical debt. The longer you're around, the more it piles up. I don't know that I could add anything more. The one thing that's a little bit unique about Skydio is that we have these hardware platforms that allow us to basically reset. The latest generation that we announced last year was three years and $80 million in the making. That allowed us to basically reset a lot. But we also have a pretty large cloud software component. There's a hell of a lot of technical debt that's been built up over the last seven, eight years. I don't have a magical answer on that one.
--:-- Host (Erin Price-Wright): Some people think to get to self-driving level four or five or robotaxi status that we don't need LiDAR. What's your take there?
--:-- Vijay Patnaik: All right, I can take that super easy question. I think the industry has somehow gotten into this "yes LiDAR, no LiDAR" binary camps. But at the end of the day, a sensor is just one other piece of technology that you incorporate to achieve the ultimate goal of level four robotaxis. If Waymo had a path to deploying the vehicles that you see on the road without a LiDAR, I guarantee you Waymo would have taken that path, because there's a lot of engineering and science invention that went into building some of the LiDARs that Waymo has. They really are cutting edge LiDARs. Waymo is not building them for the sake of, "Hey, it's fun to work on these technical challenges." There's a portion of the safety case that goes back to the question of how do you certify that these are safe enough that relies on certain properties that only LiDARs have that today the other combination of other sensors cannot.
--:-- Even in cars in China, for example, you see LiDARs in lower levels of autonomy. That allows them to bring certain capability to market faster. But also from a liability perspective, you have to really think from a manufacturer perspective. If the vehicle does get into an accident, and we will see accidents happen in the industry as the technology is maturing, would you rather have gone all the way and done whatever was possible from a safety and liability perspective to prevent that, or not? That is the question OEMs have to face when they think about whether they should put a LiDAR in a passenger car.
--:-- Ultimately, will we get to a point where systems without LiDARs are good enough to function in their respective design domain? That design domain could be mining, it could be cars, it could be some defense applications. I think in the long term, we will get there. But do I think that at the early stages of deployment of technology, taking the safest path is the right approach in order to build up trust with the technology, with residents of a city, and with regulators? I think that's a pretty reasonable approach.
--:-- Host (Erin Price-Wright): I'm super curious to hear your observations around the ecosystems you're seeing emerging. I'm sure that's also really interesting from a venture perspective. How do you see the next couple of years coming together there?
--:-- Peter Bowman-Davis: I'm happy to take a first crack at this. I think it's a really good question. It's something that we think about a lot. Historically, the industry has mostly focused around, with a few exceptions, Applied being a great example of an exception, Scale AI being another one that both businesses really took off as a result of selling software into the self-driving market. Most autonomy companies for the last decade have been very tight coupling of hardware and software, where all of the software was written almost to the firmware level for the specific hardware form factor.
--:-- What we're starting to see, partly due to the last three years of developments in AI and partly due to the broad explosion of interest in autonomy across a lot of different use cases, is more of a software tool chain emerging. There are a lot of unsolved challenges that we see pretty consistently across companies operating in autonomous spaces. Even basic assumptions about how software is built for the traditional SaaS world of the 2010s just doesn't work in the autonomy space. You often don't have Wi-Fi. You can't expect that you can just ship releases every five minutes to customers and they'll be able to update. How do you stream data from one place to another? How do you decide what decision making happens on device, on chip, on the edge versus in a cloud environment? Because it has to consider a swarm or a fleet of devices.
--:-- Those are all really interesting technical challenges and problems that I think developers will start solving more at scale in the next several years as more and more autonomy use cases come up. I imagine that the developer toolkit around building an autonomous system is going to be more decoupled from the hardware itself over the next several years. There are lots of opportunities to build new tech in those seams.
--:-- Macario Namie: From our vantage point, autonomy is what democratizes access to the drones, but it's not what people buy. They don't care about our autonomy. They frankly don't really care about our drones. What they care about is the information that the drones give them. The easier way we can get that information, the faster way we can get that information. Our next level of autonomy in the broader roadmap is going to be less about how do you fly and more about what mission are we trying to accomplish and how do you actually coordinate across multiple drones to be able to accomplish that mission in a safe way, in a fast way. That becomes much more vertically specific and use case specific. That's where the gold is and will create tremendous differentiation.
--:-- Our hope is that there is actually a very thriving drone industry in the United States to be able to combat the Chinese manufacturers. It's also extraordinarily capital intensive. That's a big barrier. We've raised a lot of money. We're the largest manufacturer in the United States, but we still have a long, long, long way to go. I hope there's a proliferation of a number of players coming in and contributing, because I think it'll be overall best for consumers.
--:-- Vijay Patnaik: In building applications for the vehicle, today, for the most part, the cars we buy in the US, in Europe, etc., they're not delightful consumer products. When you brought your first iPhone and you had a magical moment of, "Wow, this is great," some of that has tapered off at iPhone 16, which seems like the same as 15 and 14. There's really not been a great consumer product in that sense for a while. But when you go to China and you experience those cars, you get some of that "wow" and that moment of delight back.
--:-- Then you peel the layers and say, "Okay, why doesn't this exist in the cars here?" The reality is that behind your car today, there are 150 different suppliers that each provided a small ECU, a mini computer, and the OEM integrated that into a functional experience. But you almost have to redesign the car from the ground up. That's what Tesla did because they could start fresh. They could start with software engineering talent and say, "We don't need to design it the way that we did." That's a journey every single vehicle OEM is on today. That's why Applied is providing that operating system. If we can provide the operating system and give you a nice SDK to build those consumer applications on top, you can unleash your creativity and think of the car as your third space and what experiences you would want in that way. Is it theater mode? Is it something else? That's where I hope the industry goes, and I think overall it'll be just better for consumers.
--:-- Host (Erin Price-Wright): All right, that is all for today. If you did make it this far, first of all, thank you. We put a lot of thought into each of these episodes, whether it's guests, the calendar Tetris, the cycles with our amazing editor, Tommy, until the music is just right. So if you like what we've put together, consider dropping us a line at ratethispodcast.com slash A16Z. And let us know what your favorite episode is. It'll make my day, and I'm sure Tommy's too. We'll catch you on the flip side. Thank you.
