Automating Circuit Board Design Using Reinforcement Learning w Sergiy Nesterenko, Founder of Quilter on Autopilot

In this episode from the Autopilot podcast, Sergiy Nesterenko, founder of Quilter (backed by Benchmark), discusses designing PCB circuitboards end-to-end using reinforcement learning. "Autopilot" host Will Summerlin and Nesterenko cover how current PCB boards are designed and how Quilter’s tech stack enables faster board design, what better circuit boards enable in the future, their GTM and where they are seeing most market pull right now, and much more. Subscribe to Autopilot: Spotify: https://open.spotify.com/show/6YQZkKHN7EP2yWedAvSxBC?si=18377c69a2804333&nd=1&dlsi=18fee5e95b284d02 Apple: https://podcasts.apple.com/us/podcast/autopilot-with-will-summerlin/id1738163836 YouTube: https://www.youtube.com/@AutopilotwithWillSummerlin Check out Nathan's new chatbot on www.cognitiverevolution.ai SPONSORS: The Brave search API can be used to assemble a data set to train your AI models and help with retrieval augmentation at the time of inference. All while remaining affordable with developer first pricing, integrating the Brave search API into your workflow translates to more ethical data sourcing and more human representative data sets. Try the Brave search API for free for up to 2000 queries per month at https://bit.ly/BraveTCR Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off https://www.omneky.com/ Plumb is a no-code AI app builder designed for product teams who care about quality and speed. What is taking you weeks to hand-code today can be done confidently in hours. Check out https://bit.ly/PlumbTCR for early access. Head to Squad to access global engineering without the headache and at a fraction of the cost: head to https://choosesquad.com/ and mention “Turpentine” to skip the waitlist. LINKS: Quilter: https://www.quilter.ai/ Autopilot Ventures: https://www.apv.vc/ X/SOCIAL: @labenz (Nathan) @WSummerlinAI (Will) @sergiynest (Sergiy) @quilterai (Quilter) TIMESTAMPS: (00:00) Intro & Sergiy’s Background (04:31) What Sergiy learned from SpaceX (05:53) Founding thesis of Quilter and Quilter’s journey (06:57) Where would one find circuit boards? (08:11) What is the process of designing a circuit board? (09:39) Design process today with Quilter (14:34) Sponsor: Omneky (16:01) Quilter’s thesis and designing more complex circuits (18:25) How much are humans currently paid for board design (19:41) Labor dynamics in board design (20:30) Do most companies have board designers in-house? (21:14) Incentive structure (22:44) What does a high-performance circuit board look like vs low-performance? (26:37) Quilter’s technology stack (29:15) Sponsor: Plumb | Squad (31:15) How Quilter can grow with scale? (33:50) Where is circuit manufacturing happening (37:17) What other parts of knowledge work can be solved with reinforcement learning (41:03) GTM and who Quilter is selling to (42:48) Pricing (44:04) Where Quilter is seeing the most market pull right now (45:21) What makes Quilter an exciting company to work at or invest (47:14) The effects of closed research in private companies for the industry (49:25) Open source vs closed source (51:14) What Sergiy would advise to himself in his early founder (53:59) What drew Sergiy to working with Benchmark (56:19) Wrap

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

• 0:00Nathan Labenz?

  Hello and welcome to The Cognitive Revolution, where we interview visionary researchers, entrepreneurs, and builders working on the frontier of artificial intelligence. Each week, we'll explore their revolutionary ideas, and together, we'll build a picture of how AI technology will transform work, life, and society in the coming years. I'm Nathan Labenz, joined by my co-host, Erik Torenberg. Hello and welcome back to The Cognitive Revolution. Today, I'm pleased to introduce Autopilot, a new podcast hosted by Will Summerlin, which explores the past, present, and future of AI and automation with expert guests. Will is the founder and managing partner at Autopilot VC, and before that, he covered software and artificial intelligence at Ark Invest.

• 0:44Nathan Labenz?

  He was also previously the founder and CEO of Pin AI, a venture-backed startup that developed an AI-powered security platform. In this episode, Will interviews Sergey Nesterenko, the founder and CEO of Quilter, a startup that uses reinforcement learning to automate the complex process of designing printed circuit boards. As an electronics engineer at SpaceX, Sergey experienced firsthand the challenges of circuit board design. A single board can take weeks or months to design and validate. Issues and errors are common, and iteration involves the production of physical prototypes, which can be frustratingly slow. With Quilter, however, Sergey aims to make circuit board design as fast and seamless as compiling code. In this conversation, Will and Sergey dive

• 1:29Nathan Labenz?

  deep into circuit board design as it exists today, tell the founding story and technology behind Quilter, and consider both how to earn trust with AI-created designs and how to navigate the labor dynamics of an industry poised for disruption. While Sergey is quite upfront about the fact that Quilter can only handle relatively basic designs today, even this suggests major productivity gains and changes to the nature of jobs. And more importantly, if Quilter achieves its vision of superhuman circuit board design performance, which, given the success of other narrow reinforcement learning systems, seems very plausible to me, will see a much bigger impact on employment in years to come. For more fascinating looks at critical but often neglected domains and the impact

• 2:11Nathan Labenz?

  that AI automation is and will continue to make on them, subscribe to Autopilot with Will Summerlin wherever you get your podcasts. And of course, we here at the Cognitive Revolution always welcome your feedback. Right now on our website, cognitiverevolution.ai, you can chat with my personal AI clone, which is powered by Delphi AI and trained on all Cognitive Revolution episodes. Feel free to ask it any questions you have, give it a guest suggestion, or just tell us something that we got wrong. I've had a couple of good experiences with it, but I'm really curious to see how it works for others. And finally, for now, remember that I am helping a few different companies hire AI engineers. So if you're looking for a new AI engineering role, you can go ahead and DM the real me on the social media network

• 2:53Nathan Labenz?

  of your choice. With that, here's Sergey Nesterenko of Quilter on the AI Automation of Circuit Board Design from the new podcast Autopilot with Will Summerlin.

• 3:05Speaker 01

  Welcome to Autopilot, a podcast where we discuss the past, present, and future of AI and automation.

• 3:22Speaker 01

  We explore how AI is creating and disrupting large industries in conversations with top founders, investors, and historians. I'm your host, Will Summerlin. Let's dive in.

• 3:39Speaker 01

  Sergey, thank you for joining us. You are the founder and CEO of Quilter. You just raised a $10 million round led by Benchmark, so congratulations on doing that. And a little bit, we'll get into PCB design and how it's broken and how you guys are fixing it. But for context, let's start with you and the founding story. What were you doing before and why did you found Quilter?

• 4:00Sergiy Nesterenko?

  Yeah, Will, thanks for having me. It's great to be here. Yeah, so I, you know, my first job out of college was at SpaceX. I was responsible for Falcon 9 and Falcon Heavy second stage in high radiation environments. Basically, rocket, when you go to space, don't die. That was my job, which involved a whole lot of electronics. So, you know, the founding story of Quilter actually kind of was largely inspired from that, just suffering the pain of developing hardware and all the supporting hardware to develop hardware at SpaceX.

• 4:26Speaker 01

  If we, if we go back to SpaceX for a second, we'll, we'll get into what you're working on now.

• 4:32Speaker 01

  What, what lessons did you learn? Like SpaceX is obviously, you know, an incredible company, solves really hard problems. Seems to have a relatively intense culture. I'm curious like what you took away from that culture that you're bringing over to Quilter.

• 4:42Sergiy Nesterenko?

  Yeah, I mean, you know, Will, there's a lot. I think on a daily basis, if you ask, you know, the, the awesome team that I work with, like, I'll probably cite some learning from SpaceX almost every single day. You know, off the top of my head, I think maybe the biggest one is just like, how important is to have something exciting to work on? Like, you know, the fact that, you know, at SpaceX would come in the door and it's like, all right, we're actually trying to go to Mars. Like, that's not a joke. Like, we have to actually make decisions that eventually get us there. Like, that's exciting. It makes recruiting really easy. It makes showing up at work really fun. It makes you like work hard at your current project to get another cool project in the future. And like, I've seen some of that here, right? Like, you know, the team at Quilter is awesome and everybody shows up with a ton of energy every day.

• 5:25Sergiy Nesterenko?

  We've been lucky to recruit amazing talent and there's just something wonderful about working on like something hard, something important, something cool. That's the biggest one. I don't know, it's probably dozens others.

• 5:36Speaker 01

  It's like the classic missionary versus mercenary argument, right? The missionaries will be intrinsically motivated.

• 5:45Sergiy Nesterenko?

  Yeah, yeah. I, yeah, I've never worked in a context where that wasn't the kind of the driving case, but from my experience, that certainly, certainly resonates.

• 5:53Speaker 01

  And when you started to, to think about founding Quilter, like, what was the founding thesis? What problem were you trying to attack?

• 6:00Sergiy Nesterenko?

  Yeah, so when, when we founded Quilter, it actually wasn't focused at all on circuit boards, believe it or not. So this was 2019. It was a slightly different capital environment than it is today. Kind of got the opportunity to, to meet an investor, there was Michael Deering at Harrison Metal at the time. And just talk about, you know, what it might be like to start a company, what we might do. And we first actually headed in the direction of making Low cost sensors for manufacturing. That was the first kind of idea and thesis. Spent a few months thinking about that direction. Wasn't a good idea, at least not for us. We were too late to it, I think. Shifted to doing analysis on those sensors. That turned out not to be a great idea. Then shifted to reliability analysis on circuit boards.

• 6:43Sergiy Nesterenko?

  And then I kind of started thinking like, why do analysis if I can just design it correctly in the first place? So it probably was this almost a whole year of like this wilderness period of, you know, what am I actually going to do with this company? Before we landed on creating circuit boards.

• 6:57Speaker 01

  So when we think about circuit boards, maybe we can start at like a fifth grade level and then we can dig it a little bit deeper and get more advanced. If you're explaining this to me like I'm a fifth grader, what are circuit boards? I started ripping open appliances in my house. Where would I find them?

• 7:13Sergiy Nesterenko?

  Yeah, yeah. So, first of all, like, you know, circuit boards are just an important part of electronics, which is typically the thing that connects, you know, all the different pieces together and makes them come to life, right? So a circuit board, actually, I guess I have some examples right here and then I can just physically show. So a circuit board is kind of this rigid piece of fiberglass and copper that connects a bunch of chips. So these chips on a piece of electronics are the things that perform these different functions, like do some processing, store some memory, have some power, and the circuit board glues them together, right? It's a passive element and it just is meant to connect them logically. So when you're taking something apart, you'll find them just about everywhere, right? Obviously, something like your computer or your TV is probably primarily a circuit board.

• 7:58Sergiy Nesterenko?

  You'll see it in your light bulb, you know, you'll see it in the kind of control center of your stove or whatever else. Anything that has any amount of sensing and actuation will probably have electronics and therefore circuit boards.

• 8:11Speaker 01

  Does every electronic have its own unique circuit board design? Like, I would assume that my laptop has a different circuit board design than my iPhone.

• 8:19Sergiy Nesterenko?

  Yeah, that's a really good question. In general, yes, very much so. So the thing that is kind of fixed and hard to hard to create and therefore is reused between a lot of products are the chips. Right? In general, creating a new chip is a multimillion dollar proposition. And so, you know, it's very difficult to go create a unique chip for a unique product. The circuit board is kind of the opposite, right? So you can take all of these chips that already exist, glue them together in some new way and make a new product. So yeah, definitely the board in your iPhone is different than the one in your Android, different than the one in your computer, so on and so on and so forth, with relatively few exceptions.

• 8:55Speaker 01

  Is the complexity of the circuit board sort of correlated with the sophistication of the device? So like, an iPhone would have a much more sophisticated circuit board than, you know, an alarm clock?

• 9:06Sergiy Nesterenko?

  Oh, absolutely. Yeah, like the sophistication of a circuit board or the difficulty scales with roughly two things. One is just like the combinatorial complexity. So how many things are there on that board? How densely packed are they? And then the other is kind of physics related. So does the board have a lot of, you know, very high-speed signals which are difficult to work with? Or does it have a lot of very high-power signals which create heat that you have to manage? And yes, of course, like the difference between an Nvidia video card and an alarm clock is night and day.

• 9:39Speaker 01

  That makes sense. Maybe we can shift into the design phase of this. How do you, without quilter, say we're using the traditional methods of of human labor, human knowledge, work, labor, how does one go about designing a circuit board? What am I, what would I be optimizing for if I were to design one? And what tools would I be using?

• 10:01Sergiy Nesterenko?

  Yeah, so I like to think of the process of designing a circuit board in three major steps. And I often like to make the analogy to software, because more people, I think, have written some code than have designed a circuit board, so I find that helpful. So the first step to designing a circuit board is to create what's called a circuit schematic. So if you've never seen one before, it looks a lot like a block diagram or a flowchart. So you're kind of in a relatively human readable language, trying to explain what are the inputs, outputs and function in the middle that the board will do. So you might say I have a connector, I have a processor, I have some memory, I have a radio antenna, and then I drive that antenna or something like that. You can think of that a lot like writing code, right?

• 10:44Sergiy Nesterenko?

  So writing Python, this kind of human readable representation is a lot like creating a schematic. That's the first step. The second step is to convert that schematic into kind of a set of blueprints that a manufacturer can actually use to turn into a real physical design. Once you have the schematic, you have to kind of explain to the manufacturer exactly how to arrange copper, fiberglass, solder components together to make that schematic real. And so I think of that kind of like compiling code. That's the second step. And then, of course, the last step is to take that blueprint, it's called a layout, send it to a manufacturer, manufacturer creates it, you get the board on your desk, and usually you figure out that you messed something up and you try all over again.

• 11:25Speaker 01

  How long does it normally take to go from step one to having the manufacturer actually produce something? Is that typically like a week-long period or a year-long period?

• 11:36Sergiy Nesterenko?

  Yeah, it depends very much on the board you're designing. So once you've finished the schematic, doing this process called layout, it can be a couple of hours for something really, really simple with a few components, or it can be many months. If you're thinking of like a Starlink antenna or an iPhone motherboard, you're easily looking at two, three months of manual labor to make that design happen. Ironically, the process of doing layout is much faster in a lot of cases than manufacturing itself. In many cases, especially if you're willing to pay for it, once you have a design, you can have a prototype on your desk within 24 hours, or even just a few days if you're willing to pay for fast shipping from China. But the process of doing layout varies between, typically it's a few days, but it can be a few months.

• 12:22Speaker 01

  Is there some level of validating the design before you go to manufacturing? Not only to make sure that the system works, but also to make sure that it's manufacturable?

• 12:32Sergiy Nesterenko?

  Yeah, yeah, that's a good question. So there is some amount of this, yes. So in general, basic questions about manufacturing are pretty well handled today. You know, so one of the first things that you're doing when you're creating a board is you're deciding where exactly do all of the chips go? How do I make sure that none of them are colliding? And then how do I connect all of them with copper on the board? So making the basic checks around, hey, nothing's colliding. All the chips are connected the way that your schematic said. And all of the connections are far enough apart that like the copper doesn't bleed into itself because of etching imperfections, stuff like that. That's called a design rule check or DRC, and it runs very quickly in any of these tools that you use to design boards. That's not so bad.

• 13:15Sergiy Nesterenko?

  The hard part starts to show up when you start to think about more complex interactions. So I'll give you an example. One of the difficult things to understand in circuit board design is this notion called crosstalk. So if you have two connections between chips that are not tied to each other, not actually physically touching, they can still couple a lot of energy between each other. So you'll have a signal on this wire A. That signal, because it's a wave in electromagnetics, can jump over to the wire B. And in most extreme cases, you can actually couple more energy to the other wire than the one that you intended it for. So this can very quickly ruin your board from totally, from working at all.

• 13:57Sergiy Nesterenko?

  And so you have, as a designer, you know, potentially hundreds or thousands of these possible imperfections to consider. And in general, it's very difficult to make sure you've gotten them all right. Designers typically use, you know, rules of thumb to design boards, quote unquote, correctly as they make an attempt at this. There are some simulations that you might run on cases where you know there will be issues. But there is not like a one-click button that definitely gives you a sure thumbs up that you are completely confident it's going to work. And practically speaking, it almost never does work. You almost always find something that you have to go back and redesign for.

• 14:34Speaker 03

  Hey, we'll continue our interview in a moment after a word from our sponsors.

• 14:38Speaker 01

  So if I'm going to replay this a little bit, the design process today without Quilter is almost based on like tribal knowledge that sits within a designer's head. And it's a relatively manual process to sort of solve the puzzle. What is the designer's life like with Quilter? How much of that do you guys automate today? And how do you think that's going to evolve over the next five years?

• 15:00Sergiy Nesterenko?

  Yeah, so the the principle of what we're trying to do is to completely eliminate layout altogether. We want to get to the point where doing a PCB layout is like compiling code. Where, you know, in practice, pretty much in all cases, people just click a button and it runs. Now, of course, that's much easier said than done. And so that's an easy thing to say, but a very, very difficult thing to do. So with Quilter, for the boards that are kind of in our scope today, which I could talk about in a second, that is very much the process, right? That the designer will finish their schematic. They will open up the beginnings of their board file in order to enter some constraints. So for example, you might have very physical requirements on how big the board can be,

• 15:44Sergiy Nesterenko?

  where connectors go, where buttons go, where a screen might go. And then everything else that you need to do, you simply leave off the board in the design file, upload it to Quilter, and then Quilter drives it all the way home, straight to something you could submit to a manufacturer. But of course, it should make it clear that it's limited what kinds of boards we can do, right? If you give us a computer motherboard or a video video card today, we can't do it. And so you're just not going to get a good result. You probably won't even get a completed result. But for boards that are sort of within our parameters, we've done this many times and it works just fine.

• 16:18Speaker 01

  Is the thesis that over time you'll sort of move up the stack of complexity? And get to a point where you can design a board for a laptop? Or is the market big enough at the bottom?

• 16:29Sergiy Nesterenko?

  Yeah, I mean, I think that I think that there is a significant market at the kind of bottom of the complexity chain, for sure. I mean, even quote unquote, easy boards where the physics isn't that complicated, they still can take a week to do or two weeks to do by hand. So there's still a very, very significant value to eliminating that work. And even more importantly, reducing that latency of your product iteration. But we of course want to be able to do all of them. Because we eventually can unlock the potential to design better boards than humans can. And that opens up a whole new set of considerations.

• 17:02Speaker 01

  How much are humans currently paid? Do you have any sort of rough idea of the current labor spend on on board design?

• 17:10Sergiy Nesterenko?

  Yeah, totally. So on average, this is actually a profession. So there are people who specialize in this specific task of of doing PCB layout. I believe the median salary in this profession is about a hundred thousand to a hundred ten thousand dollars per year. And it's overall, it's not just those people who do this task. You know, people who, almost anybody who designs electronics eventually has to do their own layouts. And really professional PCB layout folks are typically reserved for designs that are really complicated and really mission critical. So again, thinking back to SpaceX, right? We had, you know, a team, a few dozen people who would do PCB layout. But they could only support, you know, flight computers and, you know, the intense radios and everything like that.

• 17:52Sergiy Nesterenko?

  Even for someone like in a position, you know, like myself, doing radiation testing, we had to do our own designs because that team just simply couldn't support us. And, you know, it could easily take up half of your time designing these kinds of boards. So in total, you know, we estimate that there are, you know, i- i- it's hard to get a precise number, but, you know, twenty, thirty, forty, fifty billion dollars a year of labor being spent on this task.

• 18:18Speaker 01

  What are the labor dynamics like? Are there labor shortages? Is there an oversupply of of board designers?

• 18:25Sergiy Nesterenko?

  Yeah, I think there's actually a drastic undersupply of board designers and electrical engineers in general. It's definitely a big problem. I think what you're seeing is that, and I don't have numbers for this to make it really concrete. But I think what you're seeing is that a lot of the young engineering talent nowadays is choosing to go into software and AI and robotics. And fewer are going into electrical engineering and specifically these skills of of kind of intense PCB design. So I mean, just from my own experience in SpaceX, right? I wish we had more support for doing layout with that central layout team. Because then you don't have these folks in all these adjacent roles doing that that same task. I think there's a significant shortage of this kind of skill out there.

• 19:07Speaker 01

  SpaceX is somewhat unique in terms of vertical integration. Do most companies have designers in-house or are they third, are there third-party services that people typically rely on or companies typically rely on?

• 19:20Sergiy Nesterenko?

  Well, so I think most companies that develop hardware have their own internal electrical engineering talent. That doesn't mean that they necessarily have their own layout folks. That can depend case by case. But even a company that is, you know, five or 10 people developing IoT products will probably have a few engineers doing the PCB design, the electronics design. And either they will choose to have someone do the layouts, or they will pay someone externally to do the layouts for them.

• 19:51Speaker 01

  Got it. That makes sense. Maybe we can talk about the incentive structure a little bit. If I'm a hardware company and I need to design a circuit board, what do I care about? Do I care about the time it takes to design it, the cost to design it, the efficiency of the system? Like, what are the incentives that I really care about?

• 20:11Sergiy Nesterenko?

  Yeah, I think that that really depends. That really depends on who you are within the company and the application. So we can maybe talk about a few different cases. You know, so, you know, if you are designing an early prototype, you know, for a board, that's probably most important thing is the speed, right? What's the fastest you can get it on your desk and start playing with it? You know, and just to give some some perspective to this, for every one, you know, flight computer, you're probably going to design tens or hundreds of early prototypes, you know, validating little pieces of the design, validating memory chips and power chips, and also creating support equipment to fully test it. You know, on the exact opposite extreme, if you're Apple and you're designing a motherboard for the iPhone,

• 20:54Sergiy Nesterenko?

  you probably care most about, you know, size, performance, and maybe to some extent, cost. I mean, obviously to to an extent, cost, but maybe that's not like the thing you push down the most. So it really depends whether you're building kind of a one-off prototype, in which case I think speed is very important, and of course, it has to work. Or if you're building a mass production unit, in which case like reliability, cost, form factor, maybe even aesthetic are the most important things.

• 21:21Speaker 01

  Maybe we can double down on performance. Like, what is a what is a well-designed, high-performance circuit board look like versus a poorly-designed, low-performance?

• 21:31Sergiy Nesterenko?

  Yeah, so the simplest, the simplest way to look at this is that when you just look at the raw PCB itself, right? So when you're, when you've got a schematic and it does some function, like maybe it's an Nvidia video card, right? Some fast processor with fast memory. The job of the circuit board is just to faithfully recreate that schematic. So a poorly designed board is just going to prevent your chip and your memory from working at the speed that they could work at. What you'll find is that if you run them really, really slowly, everything is okay. But then as you turn up the speed of like the communication between the memory and the chip, eventually it starts to fail. And of course, that's terrible. You know, you don't want to be bottlenecked by the performance of your board.

• 22:13Sergiy Nesterenko?

  So there's a threshold you have to meet on how good your design is to not kind of ruin the schematic that you designed. You know, the other thing is that there's other kind of limitations. So for example, to make a product in the United States, you have to pass FCC testing. You have to make sure that your board doesn't emit radio waves that jam everything else in your house, right? So of course, if you've made a poor design, that's going to happen. And you're not going to be allowed to sell the product. But if you've made a good design, that won't happen. You know, there might be some cases where that's really the ultimate goal. Is making sure that the board itself is not the bottleneck for your schematic? That it's like the raw chips that you're working with that are the bottleneck for, for speed or power performance of your, of your design.

• 22:57Speaker 01

  And if you think about kind of the arc of using some level of AI and automation to design circuit boards. If we start as like trying to get to a point where you're, you're as good as most human designers, the average human designer. And then the end of the arc is like you're better than the best human designer. Where are we on that arc? And do you think we'll get to a point within the next five years or several, where your system is better than, than the best human designers?

• 23:22Sergiy Nesterenko?

  Yeah, I mean, frankly, I don't think we are as good as the average designer on anything kind of physics and density related. Like this problem is, is very, very difficult and I don't think we've reached that point yet. I think the only thing that we are significantly better than humans at is just time. So on a board where these physics constraints are not as significant, you know, we can do in an hour what takes someone a week to do. But of course, the interesting and difficult parts of this come up when we are as good as or better than most humans at all of the physics things. And I absolutely think that's achievable within five years.

• 23:59Speaker 01

  You mentioned earlier, like, perhaps we'll be able to do things in the future with better design circuit boards than we have today. What are those things? Like, what is sort of like the horizon of of interesting problems we can solve with with really good circuit board designs that we just don't have today?

• 24:15Sergiy Nesterenko?

  Yeah, I think like, you know, a really obvious one is is cost and size reduction. So one of the things that a designer has to consider is if they sit down to do a design, it's going to take two or three months to do. Their worst fear is being a few weeks away from completion and realizing that the design's not going to close. Like, they just simply can't get everything to kind of fit and be set up in a functional way. That would require them to go back and and kind of resize the board, for example, or add layers to the board, which could easily add more weeks or even a month or more to their cycle, which is unacceptable. So what you find people doing is making a lot of conservative assumptions. Right? You make the boards a little bit bigger than they have to be. You give them more layers

• 24:57Sergiy Nesterenko?

  than they have to have. You use, you know, thicker traces or more copper poles than you might need to use. And so that ultimately results in larger boards that have lower yields to manufacture and are therefore just more expensive. I think that's probably the most obvious thing that could be done to improve boards.

• 25:14Speaker 01

  And if we kind of dig open the hood a little bit or dig under the hood a little bit, as you guys think about building your technology stack, obviously, you're not building on top of like GPT-4. This is a different problem set. Maybe you can talk through the type of machine learning approaches you guys are taking and specifically whether or not you need a lot of proprietary data and how you go about acquiring that data.

• 25:38Sergiy Nesterenko?

  Yeah, yeah, that's a great question. So the best analogy for how we approach this problem is to look at what DeepMind did with AlphaGo. So we use a branch of machine learning called reinforcement learning, where basically you create some sort of game, you create an agent that can play that game, and you record where you create some sort of score or reward that tells the agent how well it did playing that game. So obviously, with DeepMind, they did this with with the game of Go and famously eventually defeating the world champion at that game. What we're doing is this creating, quote unquote, the game of circuit board design. We're creating agents that play that game, and then the score is physics and manufacturability. So, you know, how cheap did you make this board? Is it actually going to work? So on and so forth. That's what tells you whether you won or lost.

• 26:22Sergiy Nesterenko?

  So the nice thing about reinforcement learning is that the, you know, one of the ways to approach it is that you don't actually need any external data. This is actually something that's been tried multiple times in this space, right? A lot of people have this first principles idea of like, let's take a bunch of human design boards, show them as examples to a neural network and try to recreate them. That's not how we approach this problem and that's not what we do. In part because, well, it is actually just difficult to get to really good designs. There's not a GitHub of, you know, millions and millions and millions of designs out there. And furthermore, the best designs are locked away at, you know, Nvidia, at Apple, at SpaceX, and not in the open source.

• 27:05Sergiy Nesterenko?

  But second of all, I actually think long-term, that's the wrong approach. I think that if you focus on learning from human examples, you're likely to peak at roughly human performance, which is actually what happened to DeepMind when they first started playing Go. They, they looked at a whole bunch of human examples of moves, trained on them, queried the net to make the best human-like move, and they got to like a grand master level, but not a world champion level. The best strategies came from the network trying new things and seeing how they work and remembering them. And I think that's what will eventually lead to superhuman level board designs for, for Quilter.

• 27:41Speaker 01

  Yeah, it's like there's all this unexplored solution space outside of, of where humans have been playing. And if you can explore that, you might find much more efficient designs than, than what humans know today.

• 27:52Speaker 03

  Hey, we'll continue our interview in a moment after a word from our sponsors.

• 27:58Speaker 01

  As you think about kind of the advantages of scale in this type of business or this type of product, presumably, you know, the, the more boards you design, the better you get at designing boards and creates more customers, which creates more boards, there's this sort of compounding mode. And I've always kind of thought of compounding modes in AI as like sort of a data advantage, right? If you need proprietary data, the more customers you have, the more data you generate, et cetera. How does that play out for you guys? Is, is that compounding mode exist? And as a result of that, like, do you think you'll get harder to compete with as you scale?

• 28:32Sergiy Nesterenko?

  Yeah, I definitely think so. So one of the things that's possible for us to do, and this isn't something we currently do, but that I've personally thought about a bit, when you look at any one customer... A lot of their boards and a lot of their designs start to look really similar. So I'll give you one example, like, you know, at SpaceX, we were forced to use kind of a single, relatively small set of components in our library of designs. And it's for good reason, right? You want to make sure that each part that you're going to put on a vehicle is well screened and understood, and has a good supply chain and all that stuff. And yeah, what that kind of means is that when you go to design a new board, you're going to reach into that library for a component you've probably previously used that has already been screened, rather than the 10 other alternatives that exist on the market.

• 29:15Sergiy Nesterenko?

  I mean, even without that, even for an individual designer, you're likely to work with the tool that you already know how to use. So if you're writing code, you're going to reach for a package that you've used before. If you're designing a board, you're going to reach for a chip that you've used before. So there's a potential to make little mini networks for each customer that have gotten good at their designs and only tackle their designs to keep the data kind of safe to them. But then get kind of special precision on the kinds of stuff that they're asking for. I think the other way that Quilter can kind of grow with scale is actually by integrating with manufacturers. I think that one of the other pains that exists today in design is

• 29:58Sergiy Nesterenko?

  you actually kind of have to go and figure out who's going to manufacture your board first, and then design to their specifications. Whereas if you think about it with code, it's exactly the opposite. You write your code and then you choose at the end which processor to compile it for.

• 30:14Sergiy Nesterenko?

  And so as Quilter goes and integrates more manufacturers, gets to know manufacturers, and closes the loop, that's another place where we can benefit from scale of like serving manufacturers a lot more designs at once, helping them with their margins, building those relationships, bringing costs down, and eliminating pain for the customer to submit those designs to the manufacturers.

• 30:34Speaker 01

  If you think about the manufacturing side, is most of the manufacturing done in China today? Or is it sort of being distributed into other regions?

• 30:45Sergiy Nesterenko?

  Yeah, I think it's probably true. So by volume, there's no question about it. By number of unique designs, it's probably still true that China manufactures the most. But there is a significant amount of manufacturers here in the States that do local designs. There's kind of two reasons that folks would choose to manufacture boards in the United States are primarily two reasons. One might be you might be required to do so. So if you're SpaceX, you can't send your designs to China because of ITAR. The other is latency. So when you design, when you have somebody fab your board on sure, they can, you know, build it and drive it over the same day. And you can be working on that board. Whereas if you have to fly it over from China, then you're kind of at least limited by shipping.

• 31:32Sergiy Nesterenko?

  So I don't have an exact number for how many designs are manufactured uniquely in the States versus in China. But the biggest fab in China does about 8,000 unique designs per day.

• 31:42Speaker 01

  Wow.

• 31:42Sergiy Nesterenko?

  So it's a lot.

• 31:43Speaker 01

  That's insane. Is the manufacturing process, and I guess the sort of manufacturing layer relatively commoditized? Or do different manufacturers like have some proprietary process that is a lot cheaper or a lot more efficient?

• 31:59Sergiy Nesterenko?

  The answer is kind of both, depending on what part of the difficulty scale you're looking at. So if you're thinking of a really simple board with high margins and thick traces and everything like that, sure. Every designer can do it. And, you know, the margins aren't that high for anyone and so on and so forth. But the different fabs certainly distinguish themselves once you get into special features, right? Once you get into high-speed signals with special materials or tolerances or yield, or whether they focus on quick turn or high volume, that's where you start to see some differentiation.

• 32:33Speaker 01

  When you think about quilter and your ability to sort of automate the design phase, I would assume the process today is like you're still good at manufacturing and they ship you a physical board that you can run tests on. Will that be the case in five years? Or will you guys get good enough where you can run essentially a simulation and all the tests? And by the time you get a board back from a manufacturer, it works exactly as expected?

• 32:57Sergiy Nesterenko?

  Yeah, our goal is definitely to get the board right on the first try. I think that's absolutely achievable. I think that, you know, the answer to will this board work or not is in principle computable. I don't think it's easy. I think it's going to take time, but I think it's possible. You know, there's, all we're really talking about here is, you know, capturing the process of manufacturing, Maxwell equations and laws of thermodynamics. There's a lot to do. It's a lot of compute. It's a lot of things to think of, but it is possible. And we should do it because I think that the problem with waiting for a board to come back, getting on your desk, finding an issue, fixing it and resubmitting, that's the thing that takes the most time away from a designer.

• 33:40Sergiy Nesterenko?

  You know, with code, it's relatively easy to write a bunch of print statements and debug and keep rerunning it. With a board, it's actually very, very hard to figure out what you did wrong. And that can take weeks or months, or some boards will ship with an issue never figured out. So I think there's a lot of value in getting it right the first time if you actually can.

• 34:00Speaker 01

  And you know, you mentioned compute. Our scaling laws here are sort of the same as they are in LLMs. We're like, the bigger the model, the more compute you use to train it, generally the better the performance?

• 34:12Sergiy Nesterenko?

  Certainly, increasing model sizes, increasing compute creates a better performance, but it's not the only bottleneck. You know, I think in reinforcement learning, you have kind of three, these three totally different pieces, right? Like, what is the game that you're playing? What is the agent? And what is the score? And you can't just invest in a huge agent, but then have a very difficult to play game that's super slow to evaluate and a score that doesn't actually reflect real life. So we invest not just in, you know, more compute, more model. We invest across all of those kind of pieces.

• 34:46Speaker 01

  Have you guys been constrained in any way by GPU access? Or is that not really been an issue?

• 34:53Sergiy Nesterenko?

  Not at this point. You know, we, we try to focus on problems where it's not requiring some, you know, massive million GPU for three months kind of run or problem. I mean, that's very expensive, but also very, very slow to iterate on. There's a surprising amount that you can do in reinforcement learning with a single GPU in a few minutes, right? Like, one of my favorite demonstrations to show people is that you can train a robot to walk in a simulation in five minutes on a single GPU. And so, given that's...

• 35:25Speaker 01

  Five minutes of training compute?

• 35:28Sergiy Nesterenko?

  Yeah, yep. Yeah, this is, this is, this is worth talking about. There's a, there's a cool library called JAX, I think coming out of Google, that basically gives you a whole bunch of CUDA accesses for, for training models, for building models, and so on and so forth. And what they've done is they've written a physics engine in that, you know, language, so to speak, that basically also evaluates in GPU. So you can simulate like a humanoid, or a hexapod, or a cheetah, or whatever, some sort of robot system that needs to walk on a GPU, and you can evaluate about 50 million of those episodes, so like taking a thousand steps, roughly speaking, in a matter of a few minutes. Given that that's possible, like, we try to look for things like that as much as possible. Because

• 36:13Sergiy Nesterenko?

  then, you know, we're not just hoping for the best with a huge model, we're really getting at, like, are we using the right model? Do we have a game that's that fast to evaluate? Are we looking at the right things, and we can iterate faster, and so on and so forth?

• 36:28Speaker 01

  Outside of circuit board design, what other sort of problems, or say domains of human knowledge work, do you think could be solved or partially automated with, with reinforcement learning?

• 36:40Sergiy Nesterenko?

  Oh, that's a good question. I mean, like, most explicitly, the stuff that I've been thinking about is tangential to circuit board design. So there's two other similar problems that look almost exactly like circuit board design, and that's chiplet design, and then potentially chip design. Those are really interesting. But we can kind of classify that as like electronics design. But I'm excited for it in a lot of different cases. I mean, I'd love to see more solutions doing mechanical approaches like this. You know, when you consider like... I don't know, something simpler like let's design a fan or a heat sink, right? Like what's like the right way to arrange fins on a heat sink? That's something that could be interesting to do with like reinforced learning all the way through. Like, can we design a plane? Can we design a wing that is like more efficient and has less turbulence and whatever else?

• 37:24Sergiy Nesterenko?

  These are not something I know anything about. But it would be very, very interesting to to explore and study.

• 37:31Speaker 01

  Yeah, it seems like any any sort of like almost like design optimization or you have like well-defined constraints, like the laws of physics. You sort of create these outer bounds of constraints and explore like a wide range of solution space within that.

• 37:45Sergiy Nesterenko?

  Hundred percent.

• 37:47Speaker 01

  That makes sense. As you guys think about kind of go to market and monetizing the product, who were you typically selling to? Is there like a head of electrical engineering inside of a company or is it somebody else?

• 38:01Sergiy Nesterenko?

  Yeah, that's a really good question. So, you know, our our current go to market is is primarily driven by our technical capabilities. Right? So, you know, in the in the five-year future where we can solve any board and do a computer motherboard, sure, there's going to be a head of electrical engineering or a head of PCB design that we can directly approach who's going to see, you know, millions of dollars of value for their company for this almost immediately. But we just can't do that yet. And so we're in this trickier phase of like, where can we actually solve problems today while we build to that future? The most kind of direct, interesting segment has actually been what I think of as electrical engineering adjacent roles. So I'm thinking of folks who are

• 38:46Sergiy Nesterenko?

  they require to have a circuit board to do their job, but that is not their primary function. So, I mean, myself from SpaceX was exactly this kind of example where, you know, my primary function was to test electronics and, you know, design and predict radiation effects in flight and make sure that it was all okay, which required me to design a lot of boards. But that wasn't the main thing I was trying to do. And so for me, being able to automate that piece and just have it work, it was really attractive. And I was not designing the kinds of boards that were so complicated that it couldn't be done. We're seeing interest from folks that are, you know, firmware engineers, robotics engineers, automation engineers, test engineers. Lots of these kinds of tangential roles that are like not designing the main Apple motherboard, but are designing the things that are immediately adjacent to it.

• 39:31Speaker 01

  That makes sense. And as you think about kind of pricing and like the long-term pricing model, if you're able to do designs as good or better than, you know, humans, do you think about a pricing model that sort of reflects like the human labor costs that you're replacing? Or do you think about it in a totally different way?

• 39:50Sergiy Nesterenko?

  Yeah, 100%. So I think that we do think about it with the kind of existing way to value this work. Yeah, because there's already an existing budget item, because people are already familiar with it. So the way that this is typically done in the space today, if you were to say outsource a layout to an external fabricator, is it's done in proportion to the amount of time that it's going to take. So people will take a look at the whole board, count the number of pins or the number of connections that they need to make, and charge somewhere between $1 to $10 per pin or $1 to $10 per connection. It kind of depends on where you're doing it, which country, and so on and so forth. So that's a budget item and a pricing structure that already exists that we can directly tap into.

• 40:35Sergiy Nesterenko?

  So maybe at first we'll start out on the cheaper side of things. And then over time, as we provide more value and more simulation results, then people can, we can come up to the more expensive side of that.

• 40:46Sergiy Nesterenko?

  Yeah, that's how we're thinking about it.

• 40:48Speaker 01

  Where have you seen the most market pull so far? Is it in companies that are designing? I know you guys are right now focused sort of at the bottom of the pyramid on simpler designs. Is that typically like a consumer electronics company? Is it a defense tech company? Like where are you seeing the most market pull right now?

• 41:05Sergiy Nesterenko?

  Yeah, so the interesting thing is that it's actually less segmented by the vertical of the company and more by who at the company you're talking to. You know, so whether you're SpaceX or a consumer company, you probably have both really, really difficult intense designs and really, really easy, you know, kind of intern could do it type of designs. So, you know, that's where this kind of electrical engineering adjacent thinking came from is that we were just seeing folks who were doing these boards that didn't want to be doing them and could be better using their time somewhere else. And we see that everywhere. We see that in aerospace. We see that at, you know, these chip makers. We see that at consumer companies, you name it.

• 41:48Sergiy Nesterenko?

  Certainly, I think, you know, if we had to pick a type of company that is seeing the most interest, it's those who value speed, right? So those for whom having a board on your desk tomorrow rather than next week is critical, that's like who's approaching us the most.

• 42:05Speaker 01

  That makes complete sense. As you go about building out your own team internally, you know, obviously AI talent is, is hot right now and, and quite competitive. I, I've seen this at other places and heard this from, from other founders that it's actually easier to recruit talent if you're working on something interesting. And you brought this up at the beginning of the podcast, being missionary driven

• 42:30Speaker 01

  versus, you know, building like another chatbot for XYZ. I'm curious if you're seeing that play out in real time.

• 42:37Sergiy Nesterenko?

  Hundred percent. Yeah, yeah, without a question. I mean, I think that the things that, that make it attractive to work on something like this is, well, for one, reinforcement learning itself is a more rare application of machine learning. You know, there's probably 10 to 100 times as many applications of more traditional supervised learning in the industry than there is reinforcement learning. So immediately, experts in reinforcement learning are drawn to us because we're an example of an application of their work. And so we've seen some of that. The other thing that makes it really exciting for people is having something physical that you can imagine. Right? Like, we're not talking about, you know, like you said, like a slightly better, you know, user experience score at the end of a chat conversation. We're talking about something you can hold, something you can kind of physically work with.

• 43:23Sergiy Nesterenko?

  Like, for example, one of the goals that I have for, for the company, this might be a bit ambitious, but, but we'd like to do it. I want to be able to take this video call on a quilter-designed computer by the end of this year. And I think like

• 43:36Speaker 01

  That's awesome.

• 43:37Sergiy Nesterenko?

  Yeah, I I think should we pull that off? I mean, if not this year, we'll pull it off shortly after. But like, imagine writing some code and then eventually having a board on your desk that then you made with algorithms that then does that. Like, that is so satisfying to to think about and to do. And so certainly that draws people to to tackling this problem.

• 43:57Speaker 01

  That'll be in the computer history museum one day for the exhibit. That's no, that's that's awesome. And that's that's an incredible, incredible goal. Another question on talent, you know, it seems like the sort of trajectory for a typical research scientist focused on AI or machine learning broadly. Kind of went through undergrad, went through a PhD program, did a bunch of research and then went to, you know, one of the big tech companies and continue doing research there. And and a lot of that research was essentially academic research. So it was open source. And now it seems like that trajectory has shifted. So now people are going straight from undergrad into, you know, companies, whether that's OpenAI or or you guys or somewhere else, and doing research that sort of, you know, close within that organization.

• 44:43Speaker 01

  I'm curious if that is a risk, both in terms of like the quality of talent, but also in terms of like your reinforcement learning is great. There's a bunch of examples from DeepMind that you guys can pull on. And as companies sort of close themselves off a little bit more, that that becomes less successful. And I'm curious if if either of those two two things are kind of existential risks for the for the for the AI industry at large.

• 45:05Sergiy Nesterenko?

  I I hesitate to think that that will be. I mean, just historically speaking, when companies go and, you know, require a whole bunch of talent and and therefore inject money into a bunch of problems like, you don't end up seeing something completely isolated. You end up seeing a whole bunch of like open source examples and alternatives and and a vibrant ecosystem for that. And it's hard for me to imagine that wouldn't be the case here. Like, even if the best RL engineers are going to, I don't know, DeepMind, and DeepMind starts close source everything. You know, maybe they'll maintain an edge for for a little while, but just the raw, I think, interest, the number of people who are going to go study it, and the number of people who are gonna make an open source project to try to get a job at DeepMind is only gonna go up.

• 45:51Sergiy Nesterenko?

  So yeah, I guess it's not something I've considered as a as a potential threat.

• 45:55Speaker 01

  That that makes sense. Um, yeah, and I'm curious, like, if you look at kind of the AI landscape at a at a very high 30,000 foot level, uh, there, there is this big kind of battle between open source and closed source. Uh, I'm curious if you have an opinion on on how that's going to shake out over the next, over the next five years, whether the sort of closed source foundation models run away, or open source continues to keep up and starts to create some level of commoditization.

• 46:23Sergiy Nesterenko?

  I don't know that I have a good prediction on that, to be completely honest with you. I mean, I think, again, historically, open source has been, uh, you know, incredible and underestimated. Um, you know, I mean, like, as a good example of this, take, uh, you know, uh, Windows versus Linux. Um, you know, I mean, Windows is probably still the most popular OS for for for home use, but for production servers, Linux runs everything, even though Microsoft spent plenty of time trying to make production-grade servers and Windows Server and whatnot. Um, the thing I think that's really different here is with a neural network, like, open source shines when, uh, there's a critical mass of people interested in contributing to it, and a

• 47:05Sergiy Nesterenko?

  large number of relatively small contributions results in a really fantastic product. So you think of like VS Code, where you can have 10,000 people write a small extension, and because of that, there's a huge ecosystem of things that help you. That makes a really great product. But it's not clear that 1,000 people spending an hour a week produces a better neural network, or something like that. Um, and so that maybe is an argument against open source being quite as competitive, uh, as it has been in the past. But I don't know. I don't think I have a strong bet or opinion in one direction or the other, to be honest.

• 47:40Speaker 01

  No, that's a really good point. It's like you have these like step function improvements, like Transformer, right? Um, that sort of leads to the step function, uh, you know, increasing capabilities. And if those happen sort of within a closed environment, perhaps that's a, that's an unfair advantage. Uh, that makes a ton of sense. And we're running close on time here. Maybe if we go back and you've been building Quilter for a few years, and, and, uh, I'm sure have had a lot of experiences, both positive and negative. If you could take everything that you've learned in the process so far, go back to day one of founding the company, what advice would you give yourself?

• 48:18Sergiy Nesterenko?

  That's a good question. I mean, I wish I could just like remember everything we did wrong and then replay it like very, very specifically like, for example, let's skip the whole wilderness period of a year of exploration and let's go straight to this. But I mean, of course, that's not the intent of the question. I mean, I think honestly, probably the single most useful thing I wish I invested in more early on that maybe is helpful to other founders is accounting for the fact that building a startup is really difficult and really lonely and brutal at times. And I think that the things that have been most effective for me in dealing with that is, you know, number one, being surrounded by really great people, right? Like if you, if you have a co-founder, that's fantastic.

• 49:03Sergiy Nesterenko?

  If you have friends who are building really difficult things or building other startups, that's fantastic. You know, when you're talking to investors or partners, you should be evaluating that probably more than financial terms and whatever else, because it's going to be so important. Because at the end of the day, most founders just quit at some point. And it's, it's obvious why and just surviving that is so important. I think the other is just self-care, right? Like I, you know, going through a company is a lot more stress and kind of self-criticism than, than working at a big company, even like SpaceX. I mean, maybe like watching a rocket launch on the day of launch is, is more stressful than an average day of running a startup, but

• 49:48Sergiy Nesterenko?

  on the whole, it's no comparison. And so investing in self-care, right? Like, you know, having a routine, making sure that you're always carving out time with family, making sure that you're exercising and eating well and sleeping well. All these things are obvious, but so easy to forget when you get sucked into the work and you keep running.

• 50:08Speaker 01

  That's great advice. Yeah, I remember I had a startup in a prior life and and I was like so just focused on what we were doing. I didn't have a couch in my apartment. We just like spent all the time in the office. And finally, after like, I don't know, six months, my girlfriend came over and was like, you really need to buy a couch. I was like, I just didn't even cross my mind like, but I agree. It's certainly important to take care of yourself. And ultimately, in many ways, your job as a CEO is to influence people and make decisions, right? And if you're not, if you're not taking care of yourself, it's hard to do either of those things. That's great. You raised money from Benchmark. Benchmark is obviously a legendary firm that many founders,

• 50:50Speaker 01

  you know, want to work with and many investors want to be like, I'm curious, what drew you to Benchmark and what excites you about working with them?

• 50:59Sergiy Nesterenko?

  So I got the introduction to Benchmark through one of my existing investors who spoke very highly of them. So, you know, we're I'm working directly with Eric Vishria there who I got to meet and get to know a little bit before we before we made the commitment. I think there's two things, more than anything, that drew me to working with them. Number one was experience. I think that like, there's, you know, there's a lot to be learned from from previous failures, but there's potentially even more to be learned from previous successes. And working with someone who has seen that story of, you know, seed through public company and seen it multiple times, you know, is super attractive, right? It's the hope is that that results in avoiding

• 51:40Sergiy Nesterenko?

  lots of mistakes that are unnecessary and unforced. The second thing is, you know, as any founder should, I did a lot of diligence on Eric, both on people he recommended I speak to and people I found on my own way. And I've never heard such positive founder reference. Like, bar none, I've never done a reference call for an employee, or a coworker, a boss, or a partner, or anything that came up that positive. And so that made my decision for me really quickly.

• 52:09Speaker 01

  That's awesome. Well, congratulations on the journey so far and excited to watch it unfold. And if you have a spare laptop at the end of the year, I'll certainly take one off your hands. If people want to follow the journey, or if I'm a consumer company, or a company that needs to design electronic circuit boards, or an engineer that wants to try to find a way to work with you, how can we get in touch or follow the journey?

• 52:34Sergiy Nesterenko?

  Yeah, easiest thing is our website, so quilter.ai. If you search for quilter PCB or quilter AI on Google, I'm sure you'll find it as well. Certainly, careers page, we're open to people joining us. We're in open beta, so if you're an electrical engineer, welcome to upload signs and try it out. And I'm sure we have social links there as well for kind of updates along the way. And yeah, once Will, it probably won't be a full laptop, it'll probably be kind of a Raspberry Pi looking thing that we build, but happy to send you one once we have it.

• 53:03Speaker 01

  That's great. Well, thank you, Sergey, and really appreciate you joining.

• 53:06Sergiy Nesterenko?

  All right. Thank you, Will.

• 53:08Speaker 01

  Thanks for tuning in to the Autopilot Podcast. If you're using AI to create or disrupt a large industry, I would love to learn more about your story. Drop me a line on Twitter, or go to our website, apv.vc. If you want to support the show, the best ways are to leave a review wherever you're listening and subscribe. The Autopilot Podcast is part of Turpentine, the podcast network behind Moment of Zen, Turpentine VC, Age of Miracles, and more shows for experts by experts in tech.