Pat McGovern - Bowery Capital - Where Vertical AI Wins

The Cloud Returns Podcast covers all types of software investing, whether seed, venture capital, growth equity, private equity, debt, and even the public markets. Excited to have Pat McGovern on the show. He's a returning visitor. He's an investor at Bowery Capital. And instead of describing him, I'll let him describe himself, give you a bit of background on himself and his firm. Hey, Matt, thanks for having me on. Great to be a, a second-time Cloud Ratings guest. Have we had, have you had any other two-timers? I think some on the operating side. We do a recurring series with a, a vendor or two, and I, I forget who else. I, I wasn't prepped for that other than to say that, that you are, I believe, on the investing podcast. Possibly the first return guest. Well, it's great, great to be back. Yeah, I'm Pat McGovern. I'm a principal at Bowery Capital. Early stage B2B-focused fund, primarily investing at pre-seed. We also do some seed as well. And then within the firm, I focus primarily on vertical AI. I think pre-AI, I'd done a lot of investing and consulting around vertical SaaS, vertical marketplaces. And then, you know, post the AI boom is, you know, what, what vertical technologies can do is greatly expanded. So yeah, it's an exciting time to be investing in the space. And what are your investment criteria as it stands 2026? Yeah, I'd say we're typically investing in the first institutional round of capital. So maybe the founders have raised a couple hundred thousand from like friends and family or an accelerator, but we're usually coming in and we're trying to lead, you know, a $2, $3, $4 million financing that's going to, you know, kind of kick off the company's acceleration and Oftentimes there's some traction, but at pre-seed, it's a lot of it's inception stage looking, or they may have some pilots, some MVP built. I think nowadays the expectation is you'll have some product built given how much easier it is to build now. But we're, we're getting in, in quite early. So often there's still much to be proven out. And instead of criteria, what is your kind of overlying thesis that you're applying right now? Yeah, I think where I'm spending most of my time is kind of like the AI-native business phase. So if you think about like vertical software historically, is generally, you know, a web-based UI/UX for like keeping track of your various business processes. So your customer communications and payments and scheduling, that's all kind of been solved for to the extent that like you can vibe code that yourself if you are a small business person with cloud code and some agency, as people like to say in VC these days. But I think Where I'm placing bets is that, you know, the cost of code is going to zero or close to it. AI is going to continue to get better and more efficacious at like what it is capable of doing. So does that unlock whole new business models? You see this a lot in like the AI services side of things. You're seeing, you know, venture firms investing in what are basically like law firms or accounting firms, but they're AI native. That would have been incomprehensible, you know, pre, pre-ChatGPT. You want to use that as like a demarcation line. So yeah, a lot of it is, you know, how are you using AI to solve a real-world business problem for a set of users in a way that maybe it doesn't look like a traditional software contract? It's not maybe, you know, seat-based or shooting for the highest possible ACV. It's more outcome-driven or services-driven. So yeah, that's where we're looking. We also are also still looking at, you know, big vertical AI platforms that could become a type of platform of record for a given space. But like, a lot of industries are starting to cluster around one or two early breakouts that have raised lots of money and have lots of hype on them. So we're also looking at, if you're not going to be the dominant vertical software player for one given space, like how else can you generate enterprise value in that category? And oftentimes that looks like a services business. Got it. And I know with your newsletter, congrats on— Oh, thank you. Innovator and being a VC who has a subset. I don't think a VC wants to keep SoStack as innovative at this point, but it's nonetheless. Hey, you're, you're willing to take on the pain and I respect that. And I know in the newsletter you've mapped out 6 buckets for how vertical AI companies get built. Could you lay those out and then maybe outline a bit on top of that in terms of commentary, your thesis there? Yeah. Yeah. Happy to. Yeah. So I, I've spent, you know, a number of years looking at this space now and there's kind of certain new business models you're seeing people adopt in, you know, the AI era. So I guess one, I touched on this a little, just AI-native services, right? So you're seeing this particularly in law. So you have people that are raising venture money to compete directly with legacy service providers. So like Sequoia has backed this company Crosby that announced a significant financing a few, few months back, and like they're just competing with law firms for startup business. So their ICP is like high-growth companies. Historically, those people might have gone to a startup-focused boutique or like a solo practitioner. These are like, you know, legal projects that don't require 100 Skadden Associates or something, right? But you still want, you know, a set of eyes on it. So yeah, they're mixing, you know, attorneys they're hiring with like AI engineers, and they're just going and competing for legal spend, right? If you think about it, the legal spend bucket is a lot bigger than like the IT spend within the legal industry. So The margins historically in services businesses have been less, and that's why people said, all right, we're just going to be the IT provider for this category. But now if you can use AI to do a lot of the messy human work, that was what always made professional services margins not look like software business. Now you can. So it's like, all right, maybe that's, that's now a venture bet people are making. And that's like, that was something that did not exist pre-AI. Another one you see is people that are playing— this is kind of like the pre-AI vertical software playbook of like, we're going to be the Procore or Toast or, you know, service titan for this particular sector. Like, you have all the— a lot of last-gen players in these spaces, but then you also have a new wave of like AI-native platforms, right? So you see this— Abridge in medicine has done this very well. Voice AI-heavy product, they sell to, you know, very large hospital systems. And they've gotten in with that, but now they're offering RCM services and they're kind of like being like, all right, we're going to compete with almost like Epic or someone who was like the last gen leader, right? We're going to try to be the AI native player, but we're still like a traditional software company. So that's like another angle. So sell the big chunky vertical AI platform. Another common playbook we're seeing is like voice as a wedge. So this is for non-desk workers. We made a bet on this in a company called Anata, which is for field sellers in like CPG and pharma and med device. But you're seeing this all over of just, hey, this is an industry where people are not at their desk, they're not on computers. They probably didn't use any SaaS in like the SaaS 1.0 wave of whatever you want to call it, 2010 to like 2022. But now through like mobile devices and voice as a modality, you can bring them like AI tooling and technology where they're at. So I think that's an exciting area to be building and investing because you're getting you're not even competing with like a legacy provider. Often you're competing with no technology. So you're seeing it like in all these different categories where people were not sitting at the computer all day, they weren't able to leverage SaaS 1.0, but now they can, you know, use AI on their phone. They can talk to like a person. I think also voice, the voice experience has gotten so much better, right? Like it sounds like a human now. The latency is getting better and all that stuff's going to continue to improve just based on the progress you've seen in the voice models to date. So Voice as a wedge is definitely an interesting one and one where I'm spending a good amount of time. Two other buckets that I think are— I'm going to hold off on the sixth because this is a very long answer if I do all six, but two other buckets that are exciting in vertical AI land, at least to me. One is the business in a box plus AI. This is interrelated with an AI native services thesis. You had a lot of business in a box companies in you know, the vertical SaaS market from 2015 to 2022 that raised money. Most of them did not really have like a venture scale outcome. And a lot of the time was SMBs just don't want to pay enough for the software. And there's not enough of them in your vertical and you're just like stuck, right? You can't get past $10 million or $20 million or whatever run rate it is. So you never get to like the big IPO or even the big strategic PE exit. What's different now is you have business-in-a-box companies that are like, we'll give you the business in a box. So you have a digital place to manage your day-to-day operations. Cost of building that now is like extremely low compared to what it was, you know, 3, 4 years ago. So how do you make this exciting? What you do is you then say, all right, I'm not going to charge you a traditional ACV. I'm going to provide some service and I'm going to take some percentage of top line revenue. So you've seen Moxie do this with med spas. There's a company called Alpaca Health that was doing this with behavioral therapists for people with, you know, children mostly with autism, where they give you something to run your business. But also the really big pain point for a lot of these behavioral therapists is submitting insurance claims and getting credentialed with the various insurance payers. And like, it's a giant time suck. They are generally hiring someone to do this full-time, or they're paying like an outside consultant. So you come in, you sell the business in a box, and you say, hey, we win when you win. We're going to take some kind of percentage. We're going to do some back office task that is unpleasant. We're going to help you save on labor there, but in exchange, we're going to have more taste of the upside. So like, that's, that's an interesting play. Again, a lot of these just come back like that. The AI platforms, one, the AI native services bucket, they're all like, AI can do work now to an extent. So how can you use that to either build an entirely new business model or get much higher ACVs than software historically has been able to because of this idea that you're going to be able to capture some labor budget, which is still being proven out if you're being, if you're being honest with yourself. And then the last one I'll touch on of where I've been spending time is this idea of custom automation platforms. So for the biggest, biggest, biggest enterprises, so think like a top 5 insurance payer, so a company that's like the Fortune 50, it's going to be hard to get them to like put their whole business on like a 2-year-old vertical AI platform, right? Or a lot of times these Really old, huge businesses have like insane technical architecture and processes, and there's like, it's not a way that you can just take a piece of software and plug it into what they do. So instead you have a number of companies that have raised a lot of money to go after this. So there's a company called Distill, a company called Percepta, a General Catalyst basically incubated of ex-Palantir people, this company called BrainCo that Elon Gill and other people put a lot of money into. And these companies are saying we're going to go out and just build custom automations and basically end-of-one software for super large organizations. We're going to charge them some implementation fee and then some ongoing maintenance fee. And as we build these different automation processes, we can then have some reusable technical asset that will make this not just turn into consulting because right now it is consulting, but it's still early. But You're going to have like a toolkit similar to what Palantir has done where they have, you know, they have Gotham and they have Metropolis and they have Foundry that are kind of technical assets they can deploy to customers. All these companies are trying to build that as well. So if they can automate claims for like a top 5 insurer, they can go to the other top 5 and say, hey, we do this for someone else. Like, you know, it's going to get faster each time. You're going to have reusable components. But I think that's how AI comes to like the biggest, biggest enterprise. You're also seeing like Anthropic and OpenAI are hiring consultants to do this, right? So like Anthropic, I went to a talk with their chief customer officer a couple months ago and, you know, they have forward deployed engineering teams that are basically competing with the people in this custom automation bucket. So it'll be interesting to see, is it better to be a model provider expanding into this or a startup riding their capabilities? I think it's to be determined, but at the tippy top of the economy and the hugest organizations, you're seeing a lot more hands-on kind of FD style motions to deploy AI versus saying, hey, here, log in and, you know, upload your data and, and, and, you know, it's, it's, it's much more of a, a handholding kind of exercise. Fascinating. There's obviously a lot in all of those buckets, and I think one that's probably a bit less discussed, not quite as visible compared to so many of them, would be the, the voice-enabled. And as you said, you had a deal in that sector. Could you elaborate first a little bit on that deal? Yeah, happy to. And how that fits into the sector? And the AI voice agent thesis. Absolutely. Yeah. So we led a pre-seed round into a company called Enata. You can go to their website, enata.ai. They got a great video that explains how it all works. What they're building is a second brain for field sellers. So, you know, many people in the economy that have sales jobs where you get a company car, you've got a bunch of accounts to manage. You see this a lot in CPG, food and beverage, pharma, med device. And a handful of other categories. Those are kind of the main ones where you get this account-based selling motion. These sellers spend a lot of time, you know, taking notes in their notepad and going home, typing in the CRM. There's a lot of like data leakage during that process. Stuff doesn't always get entered and they have to write all their follow-up emails. They have to do, you know, do prep for the next day's meetings. They've got to, you know, place, coordinate samples, all kinds of parts of the job that are not super fun. I think the sellers want to be selling, not doing administrative work. So Anatta, it's basically two, two technical guys teamed up with an ex-field seller. He'd been stringing together some kind of homebrewed tools to do this, and they got together to build like a proper platform. What Anatta does is it gives every field seller effectively like superpowers. So they have this app on their phone. As they're going about their day, they can talk into it and just brain dump, you know, how the meeting went, what the person cared about, what they wanted to order, what they didn't like. And then it inputs all that data into the CRM. Writes all their follow-up emails. It, you know, integrates with the ERP to place purchase orders. So you're giving back like 2, 3 hours a day to your field sellers to focus on selling versus doing admin work. And then at like the sales org side, they're getting one, sellers that are much more effective because they have, you know, they're enabled with like AI automations that they wouldn't have built themselves. And then two, they're getting also qualitative data. So they'll sell like an insights component to like a CRO or RevOps team. And you're just going to get so much more color from people talking freely than like what they put into the CRM. I mean, we have a CRM. I don't always put the greatest entries, but I feel like if I was just rambling to it about my call, it'd be a much richer kind of dataset to work with. So that's Enata. That's like a voice AI example of, you know, there's not like a last-gen competitor these people used, right? It was basically pen and paper. So you're kind of creating a new category.. And then from that you can expand down. I don't know if they ever become like a native CRM. They probably sit on top of CRMs, but you can take on more and more of the kind of like admin back office tasks these sellers are faced with. You can also just make them better at their jobs by making them more prepared for each, you know, fall meeting and just, you know, nothing ever slips through the cracks. Kind of, you know, same, you want to see a lot of VCs using Granola for like to-do lists and follow-ups. Like, I think we're kind of heading toward a future where everything's going to be recorded. Is like one of my bets. So, you know, right now it's not recorded while the seller's out there in the field. They talk to it. I don't know if that evolves over time, but I think with like Granola and some of these other voice platforms that have really caught fire, you're realizing how much more context do you get out of like capturing a full discussion than, you know, a couple of bullet points or action items afterwards. Yeah, we're super excited about Enata. They're working with a couple large sales orgs at this point. And yeah, they're kind of working to get these folks on board and they've got people live using it in the field and Yeah, excited to see where the journey goes. But if you are running a field sales org and you're not using a tool like this, I think you're probably going to get left behind. And then back to those buckets, which I really enjoyed, and then you're writing, you have this concept of compete versus enable. Do you just want to elaborate on compete versus enable? Yeah, I think this is how I see in the era of cost of code going to zero and software, there'll be like a Cambrian explosion of software because everyone can build it. All right, so like Competing purely to have the best software in a category is tricky, right? Because like vertical AI and vertical SaaS before that, historically these are winner-ish take-all markets, right? So you'll have one or two players, maybe three, that become massive companies, and then everyone else is basically like an also-ran. Historically, vertical software's bet was like, we will enable industries to digitize and be more efficient. I think now with AI's ability to basically give you human-level intelligence and the ability of agents to take actions that have real-world impact the same way an employee would, many more startups are now trying to compete directly in a given industry. We talked about this with all these law firms that are AI-native that are launching, going after big law or boutique law. You see people doing this in accounting. There's a company called Quanta. I think they just raised $15 million. Historically, you would have sold bookkeeping software to like accountants. They're just going to startups and saying, hey, we'll be your accountant. Just like, you know, pay us X amount and you don't worry about anything. So a lot of these AI-native companies that are just competing versus enabling, they're taking some unpleasant task or task that historically was very expensive, and they're either going to take it off the hands of whoever they sell to, or they're going to do it much faster, much more cheaply. Make a service available, maybe do part of market, part of the market where it wasn't before. But I think just in general, like, we backed an insurance company recently, or insurance software company, that will probably eventually be competing directly. Right now they're selling workflow automation software, but there's a point where like maybe they become a wholesaler instead of enabling wholesalers. You're just going to see this across, I think, the whole startup economy. Because like, if you're an insurance wholesaler and you're at the 5th or 10th biggest, you're still doing like $300, $400 million a year in revenue, right? That's a massive business. Whereas you're selling SaaS to insurance wholesalers and you don't become like the guy or like the second biggest, it's kind of a bust, right? So I think there's just this huge evolution of like, let's just bring technology to all these different categories versus let's sell SaaS to that category and hope everyone clusters on our SaaS. So another example of this, Sequoia just backed a company with coverage, came out of stealth. They raised like $50 million or something and like they're just an insurance broker. They go to startups and like, we'll get you insurance. We'll get you, you know, health insurance for your employees. We'll get you like, you know, whatever kind of like malpractice insurance you need as a business. And it's not like a marketplace you log into. It's not, it's like, we're just going to do this for you. And we think that we can build a really big business with good margins, like software-esque margins, but it's not like zero, zero sum, right? Like they probably won't become the biggest insurance broker on earth. They'll still probably have a great business. Whereas if you're selling software to insurance brokers, if you don't become the biggest software vendor, you're not going to be able to go public or even have a venture-scale outcome. And then another thought on compete versus enable. All the vertical AI platforms that are selling $500K-$1 million ACV contracts— I'm thinking of Harvey and Law Firm here, it's a cheap example but it's an easy one— the bet is that they'll be able to increase ACVs over time because these law firms can either run much more leanly with the same book of clients, or they can expand and now hire more people. But in general, like, you've seen a million VC blog posts on this, you know, software spend will capture labor budget, right? So like ACVs will grow because this stuff is so effective that it's worth it to pay like a heretofore unheard amount of money for software. And it could be true, could not be true, but like, if that's true, it means competing directly will work because you'll be able to have a much bigger end market and use AI to have crazy margins relative to traditional services businesses. And if that doesn't work, all of the high ACV vertical AI platforms are going to flame out anyway because they're never going to be able to like increase ACV commensurate with the amount of money and the valuations they've raised at. I'd rather almost bet on competing directly. 'Cause in either both work or neither works, but there's a lot more positive end cases competing directly than there is becoming AI platform of record for, you know, X category. And how well does compete fit within a venture capital framework in terms of horizon capital intensity? I think that's what everyone's trying to figure out. I think definitely we haven't seen a ton of these go all the way yet, right? We've seen a lot of these vertical AI platform companies, you know, get to $100 million, $200 million in ARR in a pretty quick time frame. On the compete side, I think most of these are still like Series A, Series B, so you don't really know. I mean, some of the financings they're raising indicate that they're, they're seeing, you know, meaningful pull from the market. But I think in general, like, software is another kind of reason I think the compete angle is interesting now. It's like software is not as good a business as people thought it was, right? If you look, like, look at the multiples of public SaaS companies today, it's like 3x, right? Most of them never got to that end state where suddenly they're just like throwing off like 80% margin, 90% margin revenue and just growing steadily. Like that was kind of like a mirage that didn't really come true. I think services businesses are looking better in the sense that, you know, AI allows you to run them much more leanly. Maybe you don't get to a software margin, but software margins are kind of getting compressed anyway between a lot of the late-stage public companies that just can't make it work, it seems like, in terms of business model, and then newer companies that are growing really fast. They're spending a lot of money on inference, so their margins are compressed too. They're not having the same margin you saw in 2020, let's say, because the cost of tokens and the cost of— you need the best models to compete. That stuff adds up a lot more than, you know, simple hosting costs for like last-gen SaaS where it was just like, all right, I give you a login, use it a little more or whatever, but margins are kind of fixed here. It's like there's a lot more inputs in terms of what model providers are charging you and is also your users consume a lot of tokens. So it's like a different paradigm. And then a bit of a devil's advocate on this compete thesis is you have these— let's use accounting, right? Accounting firm that's been around 20 or 40 years, regional presence, certain degree of expertise, Anthropic very well might come out, send some forward-deployed engineers for them to get better at AI-enabled accounting. Or you can contract that service if you're making a big bet on your firm becoming more AI-enabled instead of giving way to the Harvey of the accounting sector, that you can just do that and you have expertise, brand, other advantages that a nascent AI-driven accounting services company might not have. How do you respond to that? So I'd say like you see that a lot in these roll-ups where you're seeing PE— or not PE, but like growth investors. So like some Bessemer's growth fund has bought some accounting firms, General Catalyst has bought some accounting firms. Where they go and buy these mid-market accounting firms and then they are going to infuse them with AI. But I, I'm just kind of skeptical of like the organizational change that requires is not like a skill set most VC firms necessarily have a deep bench of. I think also there's like everyone's aware of this arbitrage now of like going after trying to buy these sleepy firms and mash a few together so you can buy their customer book and then Cut personnel costs, roll out a bunch of AI tools, but like the cost has gone up of buying these legacy firms. So as an investor, I think you're looking at more like a PE kind of style return on that, like 3 to 5x. You're not going to get the like 10x, 20x, 30x that like a pre-seed or seed fund is shooting for in terms of just in the most optimistic end state, where do we land? So I don't really view them as like something we're trying to bet on. And I'm also just skeptical of adoption. I don't know. I think a lot of these services industry businesses— like, I came from a professional services background. I still have a lot of friends in that world. Extremely risk-averse, extremely tough internal IT dynamics in terms of getting like next-gen products deployed. I just would rather bet on something net new than trying to buy a company and completely change the way it works, right? Like, venture generally is betting on the net new, not betting on organizational change. So that's, that's where I come down on it. And for you as a pre-seed investor, and what do you look for in a founder who's doing something net new, something that's worth you guys being in at the pre-seed that's going to innovate, take on a market scale? What's your ideal founder type for that setup? Yeah, I mean, I think I have— my views have changed somewhat in the last couple of years. I think for a long time, took in vertical software, now vertical AI as it's known, I was very much, they have to be from industry, otherwise it won't work. There's too much nuance in XYZ category that it's just not possible to do it as an outsider. I think you've seen in the AI era lots of outsiders or people that had very thin experience in an industry, maybe they worked there for 6 months or a year, go and build what are currently some of the generationally defining V1.0 vertical AI companies. So I think I index more now on technical proficiency and just understanding of how to orchestrate different foundation models and AI tools to run the business as effectively as possible, provide output that is not just AI slop but is B2B ready, for lack of a better term. So I think technical ability matters more to me now than I think it did a few years ago when stuff wasn't changing so fast and it didn't feel like you had to constantly be on the cutting, cutting edge. I think if they're not from the industry, I want to, I want to see them be like a student of the industry. I think that matters a lot to me because like, I'll look at people that are building in a category they're not from. Then a lot of times we're on calls and you ask someone, you know, what's the revenue of like this legacy competitor? And they have no idea. Or like they don't know the competitor. You're like, all right, this is like you haven't really done the work to have a right to like raise money in this category. Like, I don't care if you're not from it, but you should at least know more than I do as someone who did like an hour of research before the call. I'll interject in there. Yeah, the technical proficiency surprised me a bit because in another world you'd hear all of these AI developments, coding is getting easier, that a non-technical founder would actually have a level playing field. But I can also see the controlling for AI slop, maximize— like, there are so many innovations to avail yourself of, and that you might get even more leverage from a technical founder. But I found that a bit of a surprise, and I, I can see both sides of it. Yeah, also, we back both kinds. Like, there's no, like, set rules in this industry, but I think When I say technical, I don't necessarily mean like, oh, they're a great coder or they were like CTO of a large organization. I'm viewing it more as like, do they know what the best voice AI model to deploy is in terms of cost versus quality trade-offs? They understand token economics, they understand how to string these systems together or build good evals to actually know if the output they're selling to customers is good enough. I think that's— and that's still real. That's a real problem. This stuff changes so fast. He generally said we were somewhat technically inclined, but I think also a lot of this AI stuff's really new. Even if you were technical, you probably weren't working with transformer models and agents didn't really work or exist. There is an opening here where this whole technology shift has happened. Some people have been following this stuff for years and most people are following it for 2 or 3 years. And also, the world is like uniquely self-teachable at the current moment. So you have like infinite tutoring courtesy of Claude. And podcasts like this training. Podcasts like this. Right? Claude doesn't come up with this by itself. It regurgitates your thesis to some VC. I'm saying more for learning a new space, right? Like, I don't have a super technical background, but you, you can teach yourself a lot more now than I think you could have 2 years ago, 3 years ago, 4 years ago on the technical side of things. Yeah, it really is infinite. And I think to your point about someone coming to raise and they don't really know much about the competitors or the facts is, you know, the bar was already high, right, to succeed and raise capital. But now with all of these availabilities, like if you don't have the intellectual curiosity You're just not going to get much done, right, compared to, you know, for what you're doing. I was recently advising a startup, and, you know, sometimes you see these CARTA-type reports and you engage in them, but you don't really focus in when you're trying to give a founder advice. And it really struck me that, like, the success rate in terms of when you defined it 18-month horizon to graduating to the next round. So that's like playing the go big, execute everything. The drop-off was in the 90 to 95% range of both the seed and the A, and that's just like a good reflection of how high the bar is to do these things well, right? There's, there's no guarantee, and if you, you know, don't have the intellectual curiosity, you don't have the drive, particularly with all of these enablers, like why should you win? Yeah, I think that's right. And also as an investor, sometimes we're investing in people we've known for a while, so it's, you know, you have more context on them. But like all these pre-seed investment processes happen fairly quickly and like you're having a serious conversation. So one of the few things you can index on is like, do they take the time to build a meaningful MVP that actually works and looks decent? Do they take the time to, you know, understand like the landscape that where they're trying to build a company. And I think, so when you're missing some of those, it's like, well, I only have, I have limited observable data and the data I have is not encouraging. So that's, that's kind of how I view it too, is it's like a litmus test for use this as a proxy for like all the other things you can't observe. One other thing I've observed, and I'm kind of a safe pair of hands, former investor, but now an analyst, publish a lot of content. People come to me with like dumb questions and the like. And one recurring thing I've noticed is there is no shortage of VC content, right? No shortage. But somehow it doesn't always get through to the founders who are fundraising, and they like to just hear from someone like you, like, what do you look for? How do you evaluate them? Like, just give them something because they'll digest a clip far different than the 30 blogs available to them. And I get it, I don't, I don't read everything that's out there. But if I heard a good anecdote from an investor at Bowery Capital about how do you assess founders, like what do you, what do you look at in these, like, you know, these pre-seed processes are relatively short, how do you look at a founder? That's like overarching thing is like how big is the vision, right? Like I think when you're looking at vertical AI and people that are building in industries that are like inherently TAM constrained to some degree. Do they seem to actually have the vision of like, I want to take this thing public, or I want to get to like a couple hundred million in revenue, which is now like the bar for success in the AI era, versus are they just like trying this out, they raise some money, see where it goes. So I think I want the, the former, not the latter. So I'm always pushing on like, yeah, you're going to get to a million in, in 6 months or whatever, or you got some roadmap and your MVP is cool and you got a pilot, but like let's play it forward like 5, 7 years. Like what's the end state? Is that something that I think is worth us, you know, partnering on for, you know, 5 years, 7 years, 10 years. These companies, these journeys, when you get on in our stage of pre-seed, it's a long, it's a long relationship. So size of vision is one thing. I think another thing is just like, can I see them growing with the company? Like I think about this a lot. It's like some people are, I've heard some VCs use this term. I think it's a little It's overhyped of like, would I quit my job to go work with them? I'm like, probably not because I work in VC and I like, I like this industry. But like a version of that I use is like, could I see them like leading the 200-person all hands in 3 years or whatever? And we're like a hypergrowth company. Usually you hear the phrase talent magnet, right? Which I think is a good, a good metaphor of like, yeah, I don't know if I'm going to go work for them or not, but like, can they assemble a team of people around them? Like, do they have the, I guess, charisma for lack of a better term? And and the network to like build a real company, right? Because I think too often it's easy to get lost as like a pretty early-stage pre-seed seed. It's like, oh, they're going to get to $2 million in a year, we're going to raise an A from some great tier one and like we killed it, right? So like, yeah, that's part of it. But can this person also like run the business in its state 5 years from now, 7 years from now? So I'm just trying to play like that forward, which is— it's not easy to predict, right? Like most people generally haven't done that before. We're usually partnering with first-time founders. But I think when you talk to them about their vision, some of that comes out of like how they see, you know, company building versus just can we ramp ARR really fast this year and raise an up round? And they're like, they're kind of two different things. And another like founder dynamic I've noticed, you know, reviewing folks' decks is you have that challenge of selling a big vision but maintaining credibility. You know, some of these things are Yeah, so ambitious, or you're getting into a territory of very, very successful competitors, or the value proposition, you know, requires different skill sets. So the short version of that question would be, how do you sell a big vision as a founder to a VC while maintaining credibility or giving them some confidence that that vision is achievable? Yeah, I mean, I think for a lot of the vertical, it's hard to have like a clean answer to this across all the different business models I've discussed. So I'll just use like the vertical AI platforms one. So these are generally people are selling a large software platform into an industry that's going to like orchestrate a bunch of AI agents, allow them to use AI in their workday. And these are like the vertical companies that raised, you know, Series B, C, D. For that one, honesty is a big thing. I think a lot of founders now are forced to exaggerate by VCs that are like, oh, if you're not at $3 million in revenue in like 6 months, you're like not worth my time. And you always see these crazy quotes going out of the bar is like $10 million in a year, otherwise it's like a waste. Like, I don't think that's necessarily true. And I think unfortunately it makes a lot of founders feel pressure to like exaggerate in a way that's unhelpful. I think ways you can show you have that really big vision in the vertical AI platform scenario is like, hey, we have 3 pilots with really large credit organizations that are paying us, you know, $30K or $50K right now. And if they convert, they'll be at, you know, $300K each, right? So you're selling like, it's meaty software, it's gonna take a while to get through these sales processes. But if you see founders already getting like that much buy-in from, you know, incumbents in their category, and you see there's another like 5,000 incumbents, you can kind of get a little more comfortable with Yeah, there's a path here. I think similarly, if you're selling more of an AI services thing, where it's more outcome driven, it's probably a lower ACV. There, it's just kind of like ramp of signups plus size of end market. I mean, a lot of what I'm doing is looking at like, at pre-seed, my two main inputs are founding team, size of end market, and then vision is kind of like, comes out of that. And then that can be like a yes or no bar. And then from there, you kind of see if you get comfortable with that particular investment. But I want to see, I want to see good founders and like a really large end market or maybe a new end market. That'd be like the med spa example of Moxie, right? Like they weren't, you know, if you can be selling software into a new category that has no competition, that's a good spot to be in. That's like a really good call out though. It's size of market and founder. Yeah, it's not rocket science, but yeah, I think it's, I think also at our stage you don't have a ton to go on, right? We're investing in pre-seed companies that usually have a few pilots. One or two paid customers. They have like part one of the platform they're building done. You know, some of the other stuff is quite ephemeral. And are there any learnings from these pilots amongst your portfolio? And you know, that's a tough one. Yeah, being a pre-seed company, what are some ways to succeed with pilots and lighthouse customers at that stage? A couple different things that I've seen. One, try to keep the pilots to the shorter side if you can. Again, what short is, is going to vary. And have some kind of like pre-discussed check-in point halfway through. It could be an NPS survey of the users. It could be number of searches run. It depends on the platform, but that way you're at least like taking the temperature throughout and it gives you opportunity to either try to convert it more quickly or course correct. I think another thing you see a lot of founders doing now is like the kind of proof of value concept. So like every— and this is tricky because like you can torture the numbers to say kind of whatever you want in terms of revenue uplift and customer savings when you start the pilot, then you got to prove it actually works. But there it's like you're trying to comp. And this is like a— I think this is still an evolving discussion and a challenge in modern, like, vertical AI platforms is just what is the right price for something, right? It's like, I'm going to— every vertical AI company is selling kind of a dual-pronged story typically of like, we're going to uplift revenue because your workers are better, faster, stronger. You know, they're now AI-enabled. They can, they can do more. And we're going to allow you to save money because you can cut costs somewhere because we've automated something that, that people were doing before. Right. So you have like the revenue uplift and the cost savings and you say, look, I'm giving you $2 million of value this year on those two prongs. So you should pay like $400K for my, my, my platform. So I think that's, that's often the, the motion you're going through, it's just hard to prove a lot of this stuff in a short time frame. So you kind of need to choose like gates for like, okay, I showed revenue uplift, I showed cost savings that are achievable in whatever time frame you have, even if they're not quite like the full potential of the platform. So maybe you, maybe you give a little bit on ACV in the first year, right? But if it's actually working, there's going to be room for uplift. I think also like referenceability and case studies are very important. So we'll sometimes have companies You know, they do something at a low ACV, but they can get a great logo and a reference and a case study, and that makes them like legible to like the other, you know, 500 prospects in their market. So, you know, view it as like a marketing expense in some sense, right? Like don't worry about squeezing every dollar you can out of your first few customers. It matters much more that they're happy and referenceable and give you like a halo of legitimacy, which I think is a thing that a lot of pre-seed and seed companies are trying to cultivate all the time, right? And that's, that's the best way to do it is say, look, happy large enterprise. So don't forego that for like an extra like $50K or something in the contract. That's really helpful advice. And then just as a catch-all, like what other advice would you have for a founder? I think being a founder, your time is like your most valuable thing, right? Like you have a million things you got to do. You got to talk to customers, you got to talk to investors, you got to recruit people. I just feel like to the extent you can— this is more possible now than before— but like build some kind of personal operating system or like, you know, try to use as many AI tools personally to free up your time. One, it's going to help you like understand what's capable, what's possible, so you can apply that to your business. And two, it's going to just give you back like some of your sanity. And three, I think investors like to see like crafty AI-native founders using these tools in like interesting, innovative ways. Like we have a portfolio company that was doing some really out there stuff with building their own agent to like find people on LinkedIn talking about their customer and then message them and write comments. I was impressed by like the, the craftiness. So I think just automating as much of yourself as you can, like gives you back time. And I think VCs like seeing that you're kind of on the cutting edge in terms of trying to apply these models to like your own life and your own business, because that's effectively what you're selling to your customers. That's a good answer. I would agree with it. I think it all comes back to intellectual curiosity and drive. And, you know, you don't have a lot of data points at the pre-seed, but that reflects on the founder. And then the other aspect of your answer is total addressable market, right? So that's been phenomenal advice. And this has been an excellent episode. I really appreciate all of the time. Where can people find you? Where do you write? Where are you online in case people want to continue to get these insights from you? Yeah, you can find me a couple different places. Find me on Twitter at PW_McGovern, and then I've got a Substack, Capital Efficient, so capitalefficient.substack.com. Usually write there every week or so about just kind of what I'm seeing in early-stage markets. Yeah, and if you want to email me and you're building something in vertical AI, you know, patrick.mcgovern@bowrecap.com. You know, always happy to talk with, with founders that are kind of starting out on the journey. Awesome. Well, this has been great. Thank you so much for the time. No, thank you. Appreciate it, Matt.