Why you can't vibe-code a factory with Julian Counihan (Schematic Ventures)

Speaker A: Welcome to Driving Alpha where we feature our friends, the leading investors who drive Alpha to deliver outperforming returns. Hi, this is Brian Smiga, uh, general partner and co founder at Alpha Partners. We are the VC's growth fund to fund pro rata rights in series B in later rounds. With me today is Julianne Counahan of Steven Schematic Ventures, San Francisco. Julian I got to meet in New York before he made the move and started Schematic. Julian's recently been named a, uh, top 100 CVC by Business Insider. So, Julian, great to see your success over the last decade.

Speaker B: Thanks, Brian, and thanks for having me on. We've known each other for over 10 years and I'm, um, excited to be on the podcast.

Speaker A: Great. You were talking about industrial tech and later physical AI before anyone else. I know there had to be some key epiphany that you had maybe over a decade ago because you started the firm in 2017. Can you describe that journey and how you got to that thesis?

Speaker B: Absolutely. So I started off as a software developer working on industrial automation systems over 20 years ago. And when I started investing in New York over 13 years ago, I recognized a white space opportunity, which was the industrial space. So it was considered at the time uninvestable by most investors. But given my background in the space, I recognized a huge and growing market opportunity. At the time, the percentage of retail spend of E Comm just kept growing, growing even through a stagnant economic period. And E Com requires much more technology, much more automation. At the same time, you had a lot of exciting advancements happening in reinforced learning for cv for robots and automation. So all those things came together to give me conviction that I saw something interesting happening and that led to the launch of Schematic Quant one.

Speaker A: That's great. And right time, right place. And you made the move from New York to San Francisco. So just starting out with that move that you made as a young man before you had your first kid, I

Speaker B: think first kid was in New York, he was born in Brooklyn. And second, uh, one in sa.

Speaker A: You brought your young child to San Francisco, started a new life. Yeah. Has that been the right move? I think probably yes. But tell me why.

Speaker B: For me, as an investor, I would say San Francisco is the place to be. So what Schematic does is we invest at the intersection of technology and industry. So from my background and years of investing in this category, I have great networks across the country among the major industrial buyers and players in this space when it comes to connecting that market to the best in technology, I feel oftentimes the best in technology is here in sf. So moving here was great for, or a great fit for the strategy of Schematic.

Speaker A: Everyone's moving back to San Francisco. The New York offices are, uh, opening satellite offices in San Francisco. So once again, you were ahead of the pack there as well.

Speaker B: Well, in 2019, it was considered. That was three or four months before COVID So SF was at the time over. It was done. Everyone was leaving. It never made sense, et cetera, et cetera. So I am ahead of the sort of pack when it comes to this cycle. But as far as that cycle goes, I was probably a little too late.

Speaker A: Anyway, congratulations on your move and on your accomplishments over the last decade. Let's come down to being a very early first check funder in companies like Flock Freight and Altana. What did you recognize about either the founders or their technology that enabled you to write that first check that you saw ahead of others?

Speaker B: I'd say for a lot of the investments we made out of fundone before the 2020 Covid period brought a lot of attention to the supply chain technology space was again, the fact that I recognized a massive, untouched market opportunity in supply chain. And with a, uh, white space opportunity, you had founders starting businesses in that category who were truly highly convicted about what they were going to be doing. So Evan and his team at Altana had a background earlier from Panjiva, to really disrupt the supply chain data sphere. And they had both the experience. They were one of one. I knew, given what Moody's, IHS and the rest of the data players in that category were doing, that there was opportunity for disruption. That all came together with our investment in Altana. We were in their first round. And then same with platform science, Airspace Technologies, Fulcrum, all part of that Fundone portfolio. You had fantastic founder market fit in the industrial space. You had passionate founders who came from the space. And so that's how we made a lot of those investments during that period.

Speaker A: So your background as a software developer still plays out in many ways, both in founder selection and underwriting. But you're also building a vibe coding course online. You're continuing to code and teach others to code. And how would you speak to that passion of yours?

Speaker B: Well, I've always loved coding, but both from, um, sort of personal and professional perspective. I started coding when I was, I don't know, maybe eight. My brother and I wrote basic code that led to visual basics that led to C, and many of the areas that I wrote code in so I've always loved coding and stepped away other than some personal projects like building a wedding app for my wedding or random CRMs over the years. Stepped away from actively getting in the code until the AI coding agents came out. And so first started playing around with cursor maybe a year and a half ago. It was okay, I'd say, but not enough to pull or draw me back in. And I think the magic moment for me was Claude in December of last year or maybe a little earlier in Q4, where. And I, uh, I, I'll never forget this. It's one of those, we all have them, those sort of special technology is magic moments for me. It's the iPhone, writing in my first Waymo and coding with cloud code where it became indistinguishable between the tech and the magic. You know, how the math works and, and you know, what's under the hood. But still how that comes together as far as a, uh, personal instinctual reaction, it feels like magic. So with that reaction I just was incredibly excited to dig back in and see what could be done. And so for the past six to seven months I have been coding almost daily. This has manifested obviously I'm doing this for Schematic Ventures, despite the fact that I'm personally super excited by. This has manifested in a few different ways. One, it's what we look for when we're investing in companies. So being able to personally understand what can be done as far as software development goes today is very informative as far as how we're picking companies, how we recognize durable edge when it comes to their underlying technology and their products. And then secondly, we're automating a lot of the automatable process here at Schematic. So this allows me to really focus on where I need to focus as far as working with our founders, working with our portfolio, our industry networks, connecting with our LPs and just how I like to describe it is our AI products here allow me to operate with the same amount of focus as a GP at a multi stage who staffed up with 3 admins and 15 associates. So vibe coding has been huge for us. It is incredibly exciting. And the last way is I've decided I love talking about it, so why not launch a little short video series, sort of cataloging some of the tips and tricks that I've learned along the way.

Speaker A: And do we find that it's schematic?

Speaker B: Yes. So our video series you, uh, can find on LinkedIn, it's listed on YouTube, we've got a GitHub that's live and we walk you through how to build a full agent.

Speaker A: That's great. That's a great way to get back and I guess you won't be staffing up too much further for your next fund.

Speaker B: I we are hiring and we will be growing the team because the early stage investing process and skill is incredibly people based. If uh, you think of all the digital filters and footprints that a founder could create online and then how that differs or translates to that in person experience, oftentimes they're correlated, but oftentimes that X factor, it doesn't speak on a LinkedIn profile or a person's career trajectory. So we are hiring. I uh, do think this is a people business and the AI coding tools that we're building will just allow the investors at Schematic to focus focus on the right things.

Speaker A: Totally agree. It's a tele relationship business. At uh, Alpha we hire for EQ as well as IQ or analytics skills and today we are hiring for Vibe coding skills as well and it lets four of our investors do the work of eight. Totally agree. Congratulations on your success. I can say you're onto Fund 3, which is a real distinction. Alpha's finally on Alpha Fund 4 4, which only one in five venture firms make. Uh, it's a fun four and I'm sure you will. What are some of the advantages you have having been first to this space that you expect is going to help you build the next great portfolio in your fund3? Will it be the founders you've backed? Will it be recycling those founders? Will it be your brand? Will it be your personal brand? What edge do you feel you bring besides your knowledge of this space to attract the best deals?

Speaker B: I'd say having been through a few cycles in hard tech is one of the biggest advantages we have today and being able to separate hype from true potential. Given the fervor and heat when it comes to physical AI, I'd say everyone now almost 100% of investors will focus on physical AI and they're turning to the category. Much of it's like an escape from AI so you can't tell in software what AI will and will not take over. Where is durable value? What are we building today that will last tomorrow? So we are going to invest in the physical world where one can't vibe code a factory or an actuator or motor. So I don't think they're investing in this category because of excitement for the category. I think it's sort of a reaction to AI. Now this has led to one of the hottest markets for hard tech that I've seen in my career. But it's not the first hype cycle that I've seen in hard tech or deep tech. So being able to in the massive sort of funnel and deal flow that we now see in this particular category, separate those that will survive this cycle and build that lasting fundamental value for the next cycle, when our pre seed investments are expected to exit, is probably one of our biggest edges here at schematic. Outside of that, I've been building networks in this space for 20 years and for folks in industry that dedication, those relationships do matter. If a new investor comes to the space, they're going to find that these are not transactional relationships. So you have to show up, you have to travel to the factories, you have to travel to the distribution centers, you have to truly be a tech and thought partner for these industry connections. And you can't just use them for ad hoc diligence or one off portfolio company introductions. So that's probably our second biggest edge here is we are truly passionate about this category and we were passionate about it when no, uh, one else was or very few other investors were. And that means something to the industry networks who are vitally important to our portfolio company.

Speaker A: So to invest in physical AI, you've got to get physical and hands on. You've got to either know the space and, or visit the factory.

Speaker B: Yes. You can't build a industrial robot in San Francisco and never visit your customer site. You can't sell a chassis, a uh, commercial vehicle chassis on the app store. So you have to get out there, you have to create these partner networks and it takes time.

Speaker A: Everyone's talking about physical AI today. And between software, hardware and the integration layers between them, do you feel your next portfolio will have a blend of those three category? And how do you see those?

Speaker B: We've always had a blend and we're not changing that for the heat of this current cycle. We evaluate or our investments, I'd say considering margins and considering any sort of inventory or manufacturing aspects to the strategy or platform. So that sort of goes into the process as to how we make each investment, but we have no sort of hard preference one way or the other. What that outside of our sort of love and passion for this industry. So what that's led to is it's always been a uh, blended hardware software portfolio where we're just backing the best teams and founders that we meet.

Speaker A: Do you see new investors moving into the space? Who are the new tourists in physical

Speaker B: AI there's an absolute ton and again I think if I would ask the average new investor in physical AI why they're investing in the space, it is a reaction to uncertainty as to what AI will or will not do in the future. So the easy path is to say we're going to invest in a physical product that one can't buy code now I love that. Generalist investors I'll never gatekeep. The more folks investing in robots and industrial automation, the cooler our future is going to be be Generalist investors bring a lot of great and fresh perspective. When you've been focused on a category for so long, oftentimes you're afraid to ask the dumb questions as to how that actuator performs or the payload of that particular robot. So generalists being part of that cap table is great. Oftentimes I find it leads to a better outcome and then the more capital here, the better we're seeing more YC teams, we're seeing more teams directly out of their PhD programs leave to start startups. So I'd say the rising tide lifts all boats.

Speaker A: Yeah, as an investor just, we just announced it in Generalist AI which is a operating system for robots.

Speaker B: I think they just launched a uh, humanoid form factor today. Right?

Speaker A: Yeah, we're Naptronic Gecko Robotics and Generalist AI and we'd love to partner with you and folks like you because you bring these companies to a series B or a C round where we can finally invest at Alpha. Uh, we're really excited to know we call you guys the black belts and so Julian's our number one black belt in industrial tech and robotics. So whenever we have a question we call Julian. Here's a question. What about the capital intensity of these businesses and how should people think, think about the significant capex to building some of the physical AI companies you're involved with.

Speaker B: Yeah, it's a great question I would say at the pre seed when it's a team and idea when we're evaluating a heavy capex company what we look for from those founders is a sophisticated understanding of the financial requirements for that business model. Where are you going to get working capital? How are you going funding finance your machines, you know and it ends up being a sophisticated layer cake of different types of financing, either asset backed or otherwise. And so that's what we look for when it comes to elements of a heavy capital intensive business really just the founder understanding of that and the required plan around that. Where we often struggle is with capital intensive program precede companies where intense inventory or upfront overhead without that financial model behind it. So I think it just underlines uh, the importance of financial planning. And if that's there, great, that's fine. I'd say we've invested in probably one of the most capital intensive businesses you could Harbinger. We led their pre seed round and it is a new class C, its electric truck, they're building factories, they're building the full chassis, battery pack, et cetera. You cannot get more capital intensive than this business outside of maybe these crazy new shipbuilding.

Speaker A: Yeah, we're in, we're in Suronic.

Speaker B: So that is more capital intensive. But at the time in 2021 I'd say harbinger was probably up there. And what that team brought to our first meeting was an insanely sophisticated multi year plan as to how they would work with their vendors, suppliers, the different types of credit they would need to back each asset on the uh, financial plan. And when we see that we're excited to invest. I'd say for anybody evaluating capital intensive, either as a founder or an investor, it just behooves you to spend a little bit more time on that financial plan than if you were building a pure play software business where your margin for error is a little bit bigger.

Speaker A: And how do you temper founders when they're in a really hot company from taking too much capital at too high evaluation, especially given their capex needs. How does that conversation go?

Speaker B: We've been through that cycle before during the ZIRP period and it's a tough conversation. I would say it's tough not to get caught, caught up in the, in the market and what you're seeing your peers raise and at those valuations. But at the same time there's a clear pattern that these are cycles and should it correct, if it ever corrects, you want to be in the right place for that cycle correction. So in 22 and 23 when that cycle corrected, many of the growth companies who had spent their 20 and 21 rounds uh, faced extreme down rounds reasons or bankruptcy. And so it was the founders who raised and had that long term perspective where they could raise at that high valuation. But understand this was perhaps uh, bringing forth the next two rounds rather than one round and planning that budget around that sort of concept. So we've had a couple companies who've done that and they're still on their 21 rounds because they understood it was at the peak of the cycle and they planned accordingly. So if you're a sophisticated founder, I don't think it's a problem. But the uh, temptation is there that you raise that round and now you've got that pressure to spend it. So you, not only you have to have that long term perspective when you raise this capital, you also have to be able to withstand the pressure of your new investors to spend and grow not just to that valuation, but multiples past it.

Speaker A: So I asked you this question three years ago about humanoid robotics and having been through a lot of hypes around PDAs, the first smartphones, Google glasses, we were a bit skeptical. And you felt that their time was a ways off. Yet any change there in your view of humanoid robotics in this market? With all the hype that's going on?

Speaker B: I haven't yet seen that magic breakthrough that we saw and have seen with the coding agents around either VLA or World Model or World Action Intelligence to show breakthrough ability in the humanoid category. The humanoids I love, I would say are probably more in the consumer category where there reduced performance expectations on the buyer. Where it's a novelty, it's cute. If it gets you a drink from the fridge, that's awesome, fantastic. But when you're operating an industrial facility where you've got high precision, extreme payload and high performance requirements, novelty does not get included in that calculation. Oftentimes you'll go for an application specific robot. We've got an investment in plus one. So I would say for humanoids, the breakthrough moment for me that I think could happen over the next year or two will not be a change in the model intelligence or the physical form factor. It's going to be finding that right application that is a perfect fit, fit for where we are today for the right combination with the frontier best intelligence and the frontier best hardware. There's gotta be some application today where current performance is a great fit for that product. So it could be something like partial delivery or elder care or a consumer product or something else like that. So if we see a breakthrough, my guess is going to be that we find the right application with what we have today.

Speaker A: I love that thinking. That's so great. So, as someone who shares a heritage with Elon Musk, I'm wearing my South Africa shirt just for you. How do you view the SpaceX accomplishment coming to the market? And um, what do you have to say about SpaceX and its prospects?

Speaker B: So heavy caveats here. I'm not a space investor, I'd like to be. So it's. We've been running the fund for over 10 years and we have been growing in experience and understanding of the Space category. I like to invest in businesses where we bring value. So we're still developing those networks and understanding sort of the commercial landscape a little bit better before we make our first investment. So that's first caveat. Uh, second, I'm not a public markets investor, so as far as their valuation versus underlying performance, I am not the person to ask on that. I am a Starlink owner, which is an incredible experience. Not quite the magic moment that I've mentioned before, but up there to get this signal in remote places on the planet. So very excited by the Starlink network. But as far as the rocket business itself, I, um, am. I'm not the person to ask about that. I am thrilled about the IPO and the return of a lot of liquidity to the private capital markets, which hopefully makes its way down to great pre Seed founders and emerging managers.

Speaker A: We all are. And yeah, that capital will be going to folks like you. Final question. As a member of the seed 100 great honor, we've been talking a lot about magic moments, and I wonder if there's a magic moment that you have in your head that you feel is on the horizon that you're willing to share and talk about. Is there an upcoming magic moment that you're waiting for that you feel is in the works in the next two years?

Speaker B: I would say right now is a magic moment. That's corny, but when you think about how amazing the coding agents are, it is really. The underlying models are great, no doubt about it. And when you see the move from one underlying model to the next, it's visible in how well you can code and what it can do. But I've also seen decreasing improvements from model Release tomorrow. Release Table 5. In the brief window we had, it was great. But I've been on 55 for the past two months and I love 5 5. And even as Fable 5 came out, I found there were tasks that I was routing back to 5.5 codecs. Or, yeah, codecs. So a lot of the magic is around the harness, and it's how we use and structure our prompts and interact with the AI models that creates that impression of magic. And so the harness is just classic deterministic coding and software architecture. And so when that harness was perfected to the level it was with the current coding agents, it created that feeling of magic. It wasn't the underlying models, it was the harness plus the models. So if you think about that in every application and every industry, everything from your CRM to your ERP to how you email, I Just think we are going to have consistent magic moments in every category and many aspects of our lives as these harnesses and the software architecture around the model for these particular applications gets perfected and put in place. You've probably seen in Gmail, now they surface auto draft replies. These are getting better and better and some of them have these little sparks of intuition that makes you smile. As far as how did they know that? How do they. They're mining your data, but the ability for them to craft a draft in your voice and turn a 10 second exercise into a quick send is a little magical. So I think we're just going to be seeing that more and more this year and next year as everyone figures out in this particular application, either this corner or this corner of personal, professional, commercial, what is the correct harness to get that magic moment using AI. So that's what I see as far as where we're, I'm excited about and potential for AI in physical, industrial and everywhere.

Speaker A: Yeah. So I guess akin to the browser and akin to the smartphone, we're living in a magic moment and it's going to unfold and we're just at the beginning, we're at the head end. And I think you're right and I think these magic moments are going to happen faster and there's going to be more of them. Would you say that's the case?

Speaker B: Yeah, 100%. The rate of product releases that I'm seeing today across every category in technology is just awesome. What they can do. Now the one question that I have and a lot of people have is what is the run rate, cost and performance? So when we remove the subsidies for these models, does that sort of begin to shrink the market size and the number of applications we see out there? So if I'm developing a free AI app to automate scheduling and my AI pricing rises to 100, $200 per month, will I continue to use that? Will this create pressure for the large AI model companies? Uh, will this create new opportunities for the open source or alternate models? That is the one other thing I'm watching here, especially as somebody who uses Claude and Codex in parallel daily. That sort of Lyft Uber surge pricing battle that I see with usage limits, with credits with overage, where they're fighting for market share and they're clearly not charging you what it is costing them.

Speaker A: Okay, so our agent coding is underwritten at the moment.

Speaker B: Yep, that's what I would say. Code while you can. Who knows when they, uh, pull the plug.

Speaker A: That's a great note. To end on Julian. So Julian, top seed investor in the United States as recognized by Business Insider. Great job on your funds 1 and 2. I'm so happy that you're going on moving on to fund three. You are really doing an amazing job in this space. Thanks for being on Driving Alpha. This has been Brian Smiga, Co Founder of Alpha partners. We're the VC's pro rata fund. We invest in Series B in later rounds and in partnership with early stage VCs like Julian when they have a significant pro rata and lack the capital they do when we share the carry.

Speaker B: Thanks for having me Brian.

Speaker A: Thanks for listening. Please like and share Driving Alpha.