The Bootstrapper's Path to $10M ARR Without Venture Capital – Andrew Maguire, Volo Ventures

I think the big test though is going to be around product quality. It's much easier to build apps than ever before, without a doubt. Five years ago, you could give a founder $20 million to hire engineers and they can go hire all the best engineers. And if they don't have great product intuition and product taste and a process for building a great product, then they're just going to light $20 million on fire. It's not going to matter. The phase that we're in now is the number of high quality apps still bottlenecked by product taste and judgment. And part of what I'm personally excited about is ways in which we can use AI to make people better at creating products.

Hello, I'm your host, David Barnard. And with me today, RevenueCat CEO, Jacob Eiting. Our guest today is Andrew Maguire, Managing Partner at Volo Ventures, an early stage venture capital firm investing in pre-seed and seed stage startups founded by exceptional builders. On the podcast, we talk with Andrew about the bootstrapped's path to 10 million in ARR, what's actually investible in consumer in 2026 and why product taste is the new bottleneck, not engineering.

Hey Andrew, thank you so much for joining us on the podcast today. Hey David, thanks for having me back. And Jacob, nice to have you. I'm also here on the podcast, David. It's good to be here. I forgot to say hi to Jacob on the last podcast, so we were just joking about that. But anyways, Andrew, super excited to have you back on. And I wanted to kick things off with just your perspective as an investor in consumer businesses. So you're probably one of the few investors I know anyway that operated a subscription app, a consumer subscription app, and is now doing venture. So just start at a super high level. How are you thinking about consumer these days and investability and things like that? And we'll kind of chase those threads.

Yeah, sure. In general, I think it's just a super exciting time because what we can do with AI is so compelling compared to what was possible even a few years ago. I think what I'm seeing from an investor perspective is obviously tons of founders are chasing the enterprise. These are big markets, big opportunities, a lot of attention is going there. But I think it's very intuitive to me and probably to you guys as well that the way that we build consumer products is going to change a ton with AI and become much more powerful, much more personalized. So I'm very excited about that. I think there's a ton of opportunity. There are definitely some things to avoid along the way depending on what kind of business you're trying to build. But generally speaking, I think it's a super exciting time for a consumer.

We were just talking before we hit record about Jacob's got an open claw set up, I have a Hermes set up. The actual act of building a consumer app these days is just light years different than it was even 12 months ago. And then that's just on the building side and then what you were alluding to the kind of experiences we can now build with AI is just a whole new ballgame. And I do feel like, gosh, we're 18 years into the app store, but it feels like a whole new beginning with these AI capabilities of building all new experiences and all new apps and all new opportunities.You Guys are publishing this data on how many apps are getting posted to the app store and obviously a lot of them are using revenue cap, which is awesome. I think the big test though is going to be around product quality. It's much easier to build apps than ever before without a doubt.

But I was sort of thinking the other day about how five years ago you could give a founder $20 million to hire engineers and they can go hire all the best engineers. And if they don't have great product intuition and product taste and a process for building a great product, then they're just going to light $20 million on fire. It's not going to matter. And so I think the phase that we're in now is the number of high quality apps still bottlenecked by product taste and judgment. And part of what I'm personally excited about is ways in which we can use AI to make people better at creating products.

Because I don't think the one shot app is going to really work long-term. It's Not Going to Rise above. It's tough to not deal in anecdotes because I think every individual's slice of what apps are you using is very hard. Are there any apps that have entered your guys' personal lives in the last 18 months that you can be like, oh, this was vibe coded? There's one that I can think of on my plug, I don't know if we've talked about on the pod yet, is ATC, which is a really cool app. If you haven't tried, it's like an iteration on watching, listening to air traffic control. It's built by a very, very small team. I know for certain it was very vibe, not vibe coded, but very AI assisted. And it's a good example of how certainly with the ambition of that project wouldn't have been achievable pre-AI and is now. But again, it's like they probably could have done that before the current world because to some degree almost the money and the engineers were as

expensive, but it was attainable versus the taste and intuition to build a really great product. I don't know if you can't necessarily go to Sandhill Road and get that. It's a more rare thing. I'm constantly going back and forth on where the actual new bottleneck is. Initially, I was not on the board of like, oh, we have a taste product understanding bottleneck. I thought that was a lot of elitism and thinking that it mattered that much. But the reason I've moved on that is that I think the noise has gone up so much and there's just so many options now that almost the only way to differentiate is by having something that stands out. One, it's a great product, but two, then to really grow your quality of your product will drive down your cost of acquisition, which will then ultimately allow you to reach more efficiently a larger scale.

Another way to think about this is if you're a pre-product market fit company, more engineers is not necessarily going to accelerate your path to product market fit. There are just so many different product decisions and design decisions and are you talking to users and are you doing all this sort of stuff? I think vibe coding is going to be very successful. And certainly if you do have good product intuition, it's going to make you so much faster. I mean, that's where I think the most leverage is right now. What I'm more skeptical of is the one shot the app type approach. I think there - You mean Just turning your OpenClaw on and being like, come up with ap ideas and publish them? You don't think that's going to make a Market winning? Build me a great personalized fitness app and here's a couple other sentences.Because there's just too many decisions to make. Well, I don't know. It's like AI's good at making decisions. Again, it's a stupid word taste.

It's just like it doesn't know what information to actually consider versus not that exists. If the context is not exquisitely prepared in text for it, it's a very hard time to understand that. Okay, so I have certainly seen AI in my work come up with novel ideas. I didn't know if that were the case 12 months ago, 24 months ago. Oh, can AI not just regurgitate things it's already seen? It does seem like at least in terms of novelness coming from the recombination of existing factors, which is a big part of creativity, but I still am not sure that it knows how to choose the best. It can come up with eight ideas. I asked app ideas yesterday from AI and I think it was ChatGPT 5.5, but honestly, I was really impressed. I was like, "Oh, two of these five are really great." They're not really great, but have potential.

I've heard a lot of app pitches in my life. Two of them I was like, "Oh, this is viable. It's not just a personal CRM or whatever." But it's still, I don't know if it has, and I do think it goes down to the drive, intuition, taste to pick one, actually drill into it, actually have conviction to... And I guess this is where the human and the exceptional human still plays a huge role, which is when the agent or the model doesn't spit out the exactly great product the first time, they have the missing factors to look beyond that to solve the bottlenecks and stuff. I think the question I'm still in some ways grappling with is just how do these competitive dynamics change what it means to even, I guess from a cash perspective, invest in the app store or in apps and consumer? And I think there's also, David, you kind of brushed over at the beginning, but there's I think a big difference between consumer and app store consumer or CSA.You were

talking about Zoe Computer and Hermes and OpenClaw and those are all consumer products, but they're not app store products necessarily. There is a bit of a difference and I think it's worth differentiating. I've seen a lot of consumer investors actually kind of pivoting away from the app store as this primary software delivery thing and sort of broadening. ChatGPT is a consumer product. It's one of the most successful consumer products in history. It is on the app store, but I would not consider it an app store story. It's in part an app store story.

When you asked what apps have you adopted recently that are vibe coded? I was flipping through the first few pages of my iPhone thinking, okay, what examples do I have? Zero vibe coded, but multiple AI assisted. But it's to your point, the three I came up with were WhisperFlow, which is more B2B, but as an incredible consumer. I'm paying 20 bucks a month for it. ChatGPT and Claude. And those three exist - Yeah, which to Say They're not Vibe coded might be wrong because I think they might actually be good examples of the initial deployment of vibe Coding. Yeah. I think it's just back on this whole vibe coding thing, I was talking about taste and judgment. I think it's also just a point of view. There are so many different little product decisions that go into making an app. So we have a portfolio company called X1, and it's basically a vibe coding app. But instead of trying to one shot the app,

what they do is they break down every component of the app creation process and they figure out what are the actual decisions that you need to make throughout the course of onboarding or throughout the course of major features or throughout the course of figuring out your brand and design. And the reason that I like this is, I mean, one, it just makes it much easier to maintain an app over time because you have these discrete modular parts of the app that you can go and adjust.

But also I think there's this opportunity to educate people on how to create products. These are the sorts of decisions that you should be thinking about. Now that we bring the cost of actually building down so low, if tools can really guide people through the critical decisions, not necessarily make them for them or maybe give them like, "Here's what we would recommend," but actually force the app creator to wrestle with a hard question around trade-offs.That I think is the path to getting from apps built faster with clawed or cursor or whatever to a vibe coded ap that a non-technical person can actually make app store ready and scalable.

Yeah, I mean we did this essentially.This has been a big part of Right Catch Shtick. It's like we're a billing infrastructure, but then this podcast, a lot of what we do is about actually how to educate folks On monetization and the practice more so than the technology, at least outwardly facing, which is true. It's unknown exactly what direction the model companies are going to take it. But so far, both OpenAI and Anthropic and to some degree most of the other ones have intentionally out of safety reasons has basically made these models very, like you mentioned, point of view, they've sort of intentionally coded out the point of view. They don't want them to have a point of view that would be unaligned. Actually, sometimes in their attempt to align or make them not have a point of view, it in fact gives them a point of view, which is hyper neutrality or an attempt at hyperneutrality. But that's sort of by design. And so will we ever

get to the point where an agent or a model isn't designed that way, is designed? I've never seen, but I've heard it, I don't know if this is true, but they say when a model comes out of pre-training, they have this pre-training step where they just gobble down the entire internet and then you have this RLHF step where they basically give it ChatGPT style questions and then they give it a reward if it answers them in a good way and then they retrain the model based on that.

So there's two phases. And I've heard when the pre-train gets done, the model just sounds kind of like, forgive the phrasing, but like a schizophrenic, somebody is just sort of babbling and crazy. So there's a lot of rawness to it. The intelligence is in there, but it's very raw. And so I think the model companies actually have a lot of control on how these things end up expressing themselves. And so long path to go like, I don't know if that's forever, if that's even a limitation that these things don't have a point of view. Now an entire model that has a single point of view about fitness apps or something doesn't seem like all that useful to the world. So I don't know if they would ever build it. With the current flavor of the way these models are built, it doesn't feel like they're going to subsume the human impetus to build things, which is where outlier consumer stuff comes from. Yeah.

The point on neutrality is interesting because it's kind of like this idea. It's like there's no such thing as objective history. Right. Yeah, it's Whatever's in the training set, which is written by the victors. Yeah, everything that's an evaluation is a point of view. I Think they've actually done a pretty good job. And I think it's actually what makes the models kind of annoying to work with sometimes is they're just so milk toast and unopinionated or anodyne is maybe the right word. I don't know. They just make these points that are kind of obvious and they feel like they've discovered something very critical, very important when you talk to them. And it's like, okay, this doesn't work. That's fine for my private conversation with Claude, but for a marketplace like an app store, people aren't going to be like, "Oh my gosh, yes, what an observation. Let me download this app and make it part of my life." It's not going to work with humans at scale.

Yeah, I think it's more about the intelligence and the context to help unlock a human point of view. I think that's sort of a way that I would think about it. And the more that we can use AI to accomplish that, then that gets interesting. And I think this is kind of the big question hanging in the air as to whether consumer mobile goes away and just gets subsumed by the big model players because there's a path that some people predict that OpenAI will just be your everything ap. And so you want to track your fitness. You say, "Hey, I'm working out and I want a good plan." And it generates your plan and it tracks your everything. And it's hard for me to take that big of a leap. But to your reinforcement learning point, Jacob, I was actually talking to somebody from one of the model companies when I was out in San Francisco for WWC and asking them why aren't the models better at animations and tastes and things like that? And can't you just train on all the apps

of the app store? And then does that mean that these models can just spit out the best apps ever? And they said something obvious in hindsight, but actually quite insightful is garbage in, garbage out. Yo can't just train on everything. You need - Yeah, There's a lot that goes into selecting what's in the training sets. And what they need, and this is what Anthropic has done so well with the taste of its model and why it's better at writing than other models and has a better personality is they have psychologists and great writers and others in that reinforcement learning step in order to help guide the model to be more personable, to be more empathetic, to write better, to all these things. And so in that conversation and just generally, I don't think it's a big focus of the model companies today to get a bunch of mobile product people into a room and have them doing RL, but it may get to the point where they start doing more of that and as they go that direction,

it will at least get better. I still will have a really hard time seeing a future of non-human in the loop building really fantastic, great apps. But I think it'll get better. Yeah. I look at some apps we've gotten to know, especially in the fitness space, like ladder, gravel. There's some really good. And I think what makes them really great, the products are good. I've used many or all of them. Different structure, different folks, they do things differently. And then that opinion is something people like, but there's a lot of brand involved too. It's like Coca-Cola. It is not Coca-Cola and Pepsi because I think there's more differentiation in these apps, but to some degree it becomes what is the association? What is the brand? Who are the people behind it? More so than exactly the features. All of these apps will tend to have a comparable set of features at the limit. I don't know if this is the right analogy,

but there's a lot of classified websites that aren't Craigslist. It doesn't have to come down to the features because Craigslist is Craigslist. There's a lot of soda companies that aren't Coca-Cola. It comes into a lot of factors besides exactly what it does. Andrew, what are you using for your fitness apps these days? I use Strava still. It's a 15-year-old App. They're older. Strava's from the Late 2000s. This is the power of utility. Strava and chess.com, I might be 40-year subscribers to those apps.

That's almost a more interesting bear case for consumer software investment and dynamism than anything else because it's like, is chess.com ever going to be replaced or need to be replaced? No. But this is the power of network effects, right? Right. I'm not putting you on the spot. Would you ever invest in a new... I've got a new chess app that I'm building that I think is going to... No, but I would be a customer of one. Invest With your time. Just not your LP's Money. I need more integration capabilities from vibe coding tools to be able to do it. I can help you out. I need to use One - Shipanapp.com. Will you be my - Explain everything. Will you be my old partner? I'll Walk you. I'll help. I'll probably work with my friend Claude, but I'm sure we can get you there.

But this brings up a great point. So network effects, that's where a ventured-backed business makes sense. But there's probably a bunch of chess apps that I would pay for as a customer that could bootstrap and build $10 million a year businesses solving some niche problem. And to me, that's an extremely exciting use case for the app store right now. Yeah. It gets to the question about capital basically and different types of capital, what the risk profile is and what the right funding mechanism is. And we've talked about this, I think probably since the beginning of Revenue Cap. When Revenue Cap started, it was much more common to see venture capital style investments into early stage apps. I don't know if you would challenge this, but it definitely feels like 10 years on it's much less common than it was.

100%. And the reason's very simple. It's a recurring revenue problem. Yeah, the return's high for most of these. Maybe to some degree, I think the expectation on outliers was higher. There were a few cases of some early... Chess.com might mean that Strava's in that example. Companies that really reached venture scale, maybe there was some overestimation on how easy that was or how common that would be. Maybe that some of them, I wouldn't say stalled out, but some of them got stuck at small public scale. I don't know.

The hundred billion dollar company that - Has Yet to exist in the consumer level. Or it's like there's the handful of TikTok and social products. Yeah. Yeah, but literally a handful. There's 10. Less than 10, five maybe. I don't know where Duolingo's market cap is today, but it's probably in the single low double digit billions, I would think. Yeah, It's Like four To five. Yeah, which is venture backable, but maybe not as easy in 2026. You went with the hundred billion dollar outcomes.

But there has never been a better time to indie developer a $10 million app, right? Yeah. What you can build, everything that RevenueCat offers on the infrastructure side and then okay, paid UA, use RC Capital for working capital. Pull forward. And build really high quality niche products. And some of them will break out and cross a 10 million a year threshold and then cross a hundred million a year threshold. Yeah. But if you think About the dynamic before, in order to build company, you needed to raise venture. And then the VCs are going to be like, how are you going to disrupt chess.com's network effects? You're not going to be able to. So then you can't raise that money. And so then you can't build - Or you Can raise a very small amount. Maybe you can get a pre-seed seed if you really try hard.

Yeah, maybe. But now there's a very clear lane to build those products super cash efficiently. And I think as the tools for making these product decisions get better, I think that's going to be really exciting. I Think this is a pattern in capitalism just generally. And we're seeing it in software, which is when a new field emerges, often the input costs are very, very high. The returns are also very, very high. So you can attract this class of capital, which venture capital, it's not a modern, modern concept. At the scale, it's very modern. But you can go back to the Dutchess India company.

These trade companies in the 1600s, it's the first kind of venture where you would raise money and go try to make a bunch of money back, which doesn't exist anymore either. Venture capital, whatever form it comes in is a transient phenomenon in an industry where it's underdeveloped, costs are high. It's very speculative. But when an industry develops and cost inputs change, it's both the reward and the requirement drift. And that capital, I'm curious, Andrew, if you agree with this, that feels like the capital, at least in the last couple years, has drifted to the AI build-out.Basically every venture capital dollar just about needs to be directly in the AI build-out or attached very direct close to that. One, that's probably where the outside returns are going to come from in the next five years. But then also to your point, the need's just not there. I don't think anybody can come to you and say, "Hey, I need

$10 million for my chess.com alternative." Because it's like, why? You actually don't need that. We've been dancing around it a bunch, but I do want to get your take, Andrew, directly on what is investible in consumer in 2026. For me, it's like you need to have a credible path to building a low churn product. So network effects, utility, in some sense these are the same. I think it's probably much easier to build a really great utility product in the age of AI because AI can use personal data so much more effectively and it can get so much better over time. So that's one lane that I think is - Which Used to be a very elusive thing.

It used to be very elusive. I mean, it was sort of talked about, but is it really - To actually build that was very challenging. Thinking about a product that is going to be much better for the user in two years or three years than it is when they first started using it, not because of the promise of more features being shipped, but because of the usage Itself. Lock-in. Yeah, usage lock-in. I was talking to somebody on Twitter about picture this. It's like a plant identifier app and somebody was talking trash. Basically the counterpoint to what you're saying, Andrew, which is why would you pay for this utility?

All it does is identify plants. But I'm a daily active. I have a big plants growing at my house that I'm monitoring. And I have years of data in this stupid litle app, but I'm like a DAU probably forever now, which your point is turn combative. Well, and think about anything around health or anything around finances. These topics that are very dynamic, there's always new data, there's always adjustments that can be made. Anything where you would say to yourself, if I had a personal full-time coach for this thing that had access to 100% of the information, that would be immensely valuable. A personal nutritionist who actually knew everything that I ate and knew everything that was in the kitchen or whatever, and then can see the record over time and learn your strengths and weaknesses and how to encourage it. The actual really deep personalization that I think before historically, I think it's felt like it was a lot of work to try to get that as a user. And even then

it wasn't that good. But I feel like now it could be really good. And so I think companies that demonstrate that and lock people in that way, the utility piece is going to be forever changed. Network effects are always going to be really powerful. They're really, really hard to create. You need to probably start with utility to get to network effects because you need some way to jumpstart the network. In the age of AI though, it's a little bit complicated because you also have inference costs. When Facebook first took off, it was like there was a lot of infrastructure costs and then those came down over time, so it's kind of comparable to that. But I suspect that it's even more expensive now. If you have a very inference heavy product that - It's easier To consume. Facebook needed that inference or needed that data center build out because they had so much users. But before you had users, the cost was low. But now it's like

even to get off the ground, the cost per user is I think much higher than it was during the Facebook time. They just had much bigger scale because it was a free product. And we don't even see the labs are just absorbing enormous amounts of that. So yeah, they're subsidizing. I mean LPs are, right? Basically the investors in those labs are eating a lot of that cost now. We're struggling, but RevenueCat is integrating a lot of AI features now. And even we're having to be like, well, there's cost decisions that go into these things. And even at our scale, especially for speculative features like, oh, would this be valuable to a user? Can we afford to do this for every user every day? And we're not talking about crazy stuff, but if we wanted to make one LLM call per app per day, that would be tens of tusands of dollars a month. That's a non-trivial investment for a small feature. But it will

Come Down. It will come down. Will it? I hope so. Yeah. It may get worse before it gets better though. Yeah, I know. Yeah. Opus has not come down in six months. And so I'm a little bit skeptical right now. Eventually - Nobody's made a comparable model to Opus. 5.5 is comparable, but it's not quite the same. So there seems to be enough differentiation that they can maintain pricing power. I had this thought the other day where I was like, because part of the discussion now is, well, use a dumber model for a simpler task and do all that sort of routing and stuff. What my experience has been when I've done that, I'm like, once you taste the good model, I can't - Yeah, because I Want to use it for everything.

I can't. Yeah, I can't. Because it will still sometimes screw up even trivial tasks. And I'm like, I don't trust the outcome or it makes mistakes. And so I do want the best model for everything, which is really expensive. But at some point it actually will become true that a more basic, cheaper model is sufficient for a lot of the stuff. I'm not sure we're actually there yet. You can use CPUs as maybe a crude analogy to this, which is where for the period of 1987 or to 2010, every CPU bump was so critical and important and you always got the fastest CPU because it really mattered. But then in the 2010s, it kind of became that all CPUs were prety much fast enough and It Became a much less interesting differentiator. So at some point we will be so saturated on other bottlenecks or whatever that it won't matter anymore. But yeah, as soon as Fable came out two weeks or whatever that was for the brief period, it was out, I upgraded all

my birth percent because I was like, "Yeah, I'll pay twice as much. I don't do that many queries." For my high intent stuff, for my open call running, I'm not going to put that on the highest and most expensive model just for price reasons. Backing up, I have to agree that price will come down. We'll build more data centers. The current technologies, we've not even gotten super efficient on how we do inference today. There's a lot of optimizations that can be made yet. So between just capacity improvements, other smart optimizations, the price and just basic economics.

Models are amazing, but they're not actually that differentiated. The only company that it seems like right now is Nvidia, TSMC and ASML are the only companies that have unreplicable technology in this stack. The models themselves seem like get a big, it's just oversimplification, but you get a big pile of data, you pre-trade over it, you RLHF it, you'll get something. Well, that's why they're so focused on finding differentiated data now. That becomes the thing.

I think you're both right that it will eventually come down, but when you're making a financial bet, you also have to be right about timing. And so if you're going with Fable, rumors are it might come back this week, and if you're building your business around Fable and it takes five years to get a Fable level model at more reasonable prices, that's a very different business outcome than a Fable level model in two years, in 18 months. And guessing at that time horizon, that's what so many people are doing right now is we're doing it at RevenueCat. We're using Opus48 for our Ricoh product. It is amazing. I've been shocked at how good it is at those things, but we're making a bet that that will get cheaper, but how quickly will it get cheaper and then how's the usage going to go and Everything's just a bet. I explicitly killed us.

We were starting to build some cost control measures and I Was like, Eh, that's what the venture capital is for, right? It's like, let's just let it rip. That's what I was just about to say.This is where venture capital is actually useful. If you are building a business with network effects, let the VCs bet on the fact that the cost of inference will come down if it's expensive, but people are added to a product that's essentially growing virally or something. That's the use case to go get the venture funding to be like, okay, we have an amazing product. People are totally addicted to it. It's growing super, super fast. However, there are these really high variable costs right now and everyone believes at some point they'll come down and as long as we're growing, keep pumping money.

Yeah, I mean this is the bet that the core model company, I mean the biggest venture round, private investment rounds of human history by far margin because of that exactly. They all understand the winnings. I mean that's the other aspect, Andrew, to the go get the venture capital if you have these cost of scaling. The eventual prize has to be big enough because these folks want to make money. I imagine this is as an investor, making money is important to you. Would you say that's fair?

Yeah, I mean LPs tend to care about that. Yeah. So LPs, that's another thing. Did you know this? I think probably everybody knows this, maybe not. So when you invest, it's all of your money, right? It's like Andrew's personal money that you invest? No. Is that not true? No? Whose money are you investing? LP's money. Yeah. LPs? What's an LP? A limited Partner. A limited partner. You raise a venture funds from people who invest in the fund.

They want to invest in startups, but they aren't experts at that. So they invest in a fund instead. The fund invests in the startups. And so it goes. I did not know that when I raised my first venture investment, by the way. I did not fully understand that dynamic. A little bit. Kind of, sort of, but didn't really understand. The LPs in 2026 are seeing the SpaceX outcomes. They're seeing the anthropic and the OpenAI outcomes. Yeah.

Well, they've seen the last three anthropic rounds and they're like, why am I doing seed investing? Why don't I just try as hard as I can and 3X my money and just mega anthropic round?That's the facts on the ground right now. I think it's probably affecting every stage of venture right now. I would say even early stage is just being like, why the hell don't I just write into the next OpenAI round? But that might be over Actually. So there's this other, I mean, this is getting a little into the weeds on venture, but there's also this multi-layer SBV thing that happened, which is like, okay, yeah, sure. I want to invest in anthropic, but turns out I don't know Dario, probably not getting on the cap table, but I can invest in some fund that invested in a fund that invested in a fund that invested in anthropic. And then each of those funds has - You learn How Fees Compound.

So they're classically happened with SpaceX. And there are a lot of people that just like their returns are diluted down to nothing or they don't even know who's managing the money above them. And it's a very dicey game to play. This is why we have these highly regulated public markets at the late stage because the Bad Behavior... No, I didn't say bad behavior, but the lack of transparency causes real risk, but not the kind of risk you should take in investing. It's A systemic - No, it's the wrong kind of Procedural risk, not actual venture risk. To this moneymaking point is that it's a point I wish I had known when I did our... I mean it ended up working out, but I had somebody slide into my DMs this year who had an ap, had a few thousand dollars in AR or MRR, was doing okay, was starting to take off and is like, "I want to raise money." I was like, "No, you do not. " And I asked them, I was just like, "Okay, so tell me how you sell this company for a billion dollars." And they're

like, this and that, coming up with extreme more grasping for straws. And I was like, "Okay, Unless that is your goal and you're going to do nothing but focus on that outcome, do not raise venture capital. All it will do is limit your options and make you miserable and you'll have mad investors. If you have good investors, it'll probably be fine. Nobody's going to jail, but it's just going to make your life hell." And luckily, I don't think somebody like that would probably accidentally raise venture. I think it's like venture capitalists are pretty good at not making those bets. They can read that. But I do think it causes a big distraction. I mean, people are obsessed with applying to YC. I don't think it's necessarily... YC's great, whatever, but they're tooled for you going for that billion, 10 billion, $100 billion outcome. It's not going to be as important or as effective for you if you're not on that trajectory. And back to your

point, Andrew, we've never been in a better time for somebody to build a company, an app company that's going to make $10 million a year. And as a founder, you talk about founder outcomes. Amazing. You've seen it. Just because your company raised money and had a bigger outcome and a higher valuation does not mean you as a founder did better. Well, even just understanding preferred versus common stock. It's like if you raise a $20 million Series A, those investors get preferred stock. If you sell your company for $20 million, they get all $20 million.

They get paid first. Investors are more protected than founders in downside scenarios. And Depending on the way it' structured, it could be a 2X liquidity advantage or a 3X liquidity where if you sell it For 40 Million - You rarely see that in 2026 unless it's Restaurants. In San Francisco VC, but in Austin VC and other markets, you do still see this kind of liquidity. For the listeners out There, If you get a term sheet with anything but a 1X preference, do yourself a favor and don't take it unless you're desperate.

At the end of the day, you should actually believe that you can build a multi-billion dollar company to go raise venture because there's another credible path now because you don't need to go hire a bunch of engineers. You can vibe code it, be really thoughtful about your product, and you can build a multimillion dollar a year business with no venture, own the whole thing, and that's awesome. And you have to respect what type of business it is. It's not just what do you hope it could be? It's like some companies have the space to become massive if you execute really well. Others you can execute incredibly well and they're just still not going to get that big. So just align the capital strategy to reflect what type of company you're trying to build.

And include not just cash capital, but your time and not just your hours, but the years of your life. If you're going to spend five, 10 years on something or two years on something, know what your possible best outcome is and make sure that's aligned with. I won't disclose too much, but I just spoke with Curtis from Schlups who've had on the podcast and famously bootstrapped. I think they're, I don't know, 12 people now. I don't know. The company's in every year he just reinvests and keeps going and it's like he's built a really amazing business and he's very happy with how it is. He's built a great product with a good team and they have space to build and probably maybe could have been a venture thing if he wanted to do that, but was very conscious. He did a very thoughtful and conscious way of choosing how to build that business.

That's the kind of thoughtful decision-making people should be making. To David's point, it can change later. You can build a $10 million business and then realize, wow, we've earned some insight into where the multi-billion dollar company actually is. Now we'll go raise to tackle that. And you're at a higher Valuation, less solution. Obviously your terms are going to be awesome. But I think there are other cases where it's like, Hey, we're going to swing. We're swinging for some massive, massive, massive thing. We know the market is huge. We know it's going to be expensive to build. And then sure, go raise venture.

Is there a middle path Or should there be a middle path though where, I mean, even with my little vibe coded app that I've been working on the last three months, I've probably spent a thousand plus dollars on my $200 anthropic subscription, my $200 ChatGPT subscription, and I put $1,000 into UGC and I wasn't expecting a big outcome on that, but at some point you do need some money. These things aren't free. To get the ball rolling with scaling, you probably do need 10, $20,000 to be investing. And that's on the way low end. If you're going to try and scale paid ads, you need probably a couple hundred thousand. Is there a middle road for products that are showing traction but aren't venture scale but can move fast? Curtis, that's a great example. Could $300,000 four years ago helped him shortcut two years of slow growth and pain? And should there be a product for that? What's the middle road?

I think that there's an opportunity for a new venture lane to emerge that takes advantage of what you're describing. Or just a new classic capital necessary Venture. And I think it makes a lot of sense because I think it is going to be a lot easier for a bunch of reasons for talented people to build these multimillion dollar a year apps. And maybe it's something like small checks with a profit share. The idea is to build a cashflowing business and venture is built around the power law. So this would not be a power law driving at all.

You have to finance your nine losses for your one win. That's why venture needs those outcomes because five go out of business, four 1X, and then one returns all the money is typically it. You Need to make outliers. But I suspect that there's a strategy around generating seeding really strong developers that generate cash flowing businesses that aren't going to require additional capital or that's not the plan anyway. The Funny thing is those existed and all the ones I'm aware of actually stopped that kind of investment in the last two to three years. Tiny funds, NDVC, ComFund, three off the top of my head that were doing almost exactly what you're saying. And maybe they just bailed at the exact moment when that curve would've bent the right direction for them. I don't Know. I don't know. I'd be curious to answer things, but in an era where the S&P has been just on a generational rip, it's really hard to get capital to go into anything but treasuries

and the S&P. You know what I mean? Why would you bother? And it's not just the return expectations, it's like how long do you take to get the return? There's also the risk, but so you get the risk right. Managing private investments is just a pain compared to just having your money in your Robinhood or your Fidelity or whatever and having some very basic money investments. And that's what you're competing with. And even a good venture fund, a great venture fund, they 5X the fund in five years, something like that. That's an amazing...

But most venture funds aren't, those are the outlier returns. I think in venture, if you can beat the risk-free rate of return, you're at least above water. And a lot of them are probably close to that. And so it's just hard. Those dollars are fungible and you're competing unless you just really believe in the thesis, you like the idea of it, you're advocacy investing. You know what I mean? You're investing in something because you believe in it, not for purely capitalist reasons. But at scale, capital will just flow where capital has grown the best. And so these things have to really compete. But I'm with you, Andrew. It feels like, and I don't know what the structure looks like. We have debt.

Are we looking at specialty lenders that understand these businesses better? Maybe it's the RC venture arm. Yeah. Well, I mean, we're talking about this now. We have Daily Payouts, which is a factoring product. In theory, we're not far from being able to build debt-based products where we loan against your current run rate. But the thing with debt is it's debt. You're borrowing against your future. I guess there' two differences between debt and equity, but debt, you're borrowing against your future, which is kind of true for equity as well. When you sell equity, it's like, hey, you're selling a share of all future profits. When you sell equity, you're cutting people into a share of all future profits. The difference with debt is the profit's capped that you have to give out, but they also have downside capture. So they get your business if it goes the other way versus equity is like the upside's uncapped, but the downside

is capped. And there are combination products. Venture debt is a combination product where it's debt, but it also includes some warrants and things like this on equity. There's no free lunch. If something is a good return on capital, it will exist as a product probably eventually. If something's not necessarily a good return on capital, maybe it won't, would be my guess. But yeah, I mean, why would something like a RevenueCat like us falls in that specialty lender, specialty financier? And industries have these. I'm not certain of this, but I'm sure oil and gas has specialty banking that understands their specific needs. I'm sure just name your industry there's a specialty lender or banking or investment arm that helps understands these industries better and can underwrite them. It's even in venture. Venture funds have their focuses because they can understand those businesses a little bit better and make better bets. It

Could also be just a studio that adds a lot of value too. Well, Actually, Bending Spoons is maybe a great example of this, but that's later stage. It's financial engineering, but plus operational engineering, it's halfway between a private equity fund roll up and an operating company. But yeah, maybe that exists at the earlier stage. I don't know. There is an interesting financing product that I became aware of just in the last year when Ladder used it is General Catalyst Growth Fund. That's a really fascinating model. You've looked into that, right, Jacob? So it's essentially debt, but it's like user acquisition financing where they fund up to 80% of your acquisition. And then structurally, the way it works is they do it based on cohorts. So if in January you spend $10 million, they'll finance eight million of that and then they track that January cohort and they get repaid. In their case,

I think it's only a 10% annualized interest, they get paid back out of that cohort's earnings. If that cohort takes three years to fully pay it back with the 10% interest, it takes three years to pay it back. The nice thing about that is one, it's not dilutive funding. And two, it's not a recourse loan where they're going to come take your anniversary or your - It's not actually Debt in that case. It's not an indie putting it on their credit card where if they get too far upside down on their credit card, they're going to be in trouble. And so that is interesting. That one is only available to much bigger apps, much more proven, much more. But there's probably room for that type of a product to exist on a smaller scale as well. I Would be really curious to know, and this is without any insider knowledge, again, it's 10%. It's like, why would they get through all that trouble when the S&P's making nearly that? You know what I mean? You got to set that up.

There are some risks involved. They have to finance the losses in that interest rate. Well, they set up that fund when interest rates were only 3% and now it's more like 6%. So things look different. Cost of capital is different. Return on investment is different. General catalyst. Yeah. Yeah. So it's portfolio value add to win deals. Oh, On the equity side. Which then in that case, it's a hybrid product like equity plus debt.

It's almost Like a venture venture debt style. I mean, I'm not familiar with it, but that's what would make sense to Me. Yeah, that's what I mean. These things all sort of flow to what's the risk-free rate of return given a time and the interest rate. This really affect these things, which when you're starting out, David Davis, when you were talking about, oh, I put a couple thousand into this, a couple thousand into that. You have the benefit of having seen scaling an app up and understanding the cashflows. There will always be an initial capital investment needed to get something off the ground. It is, I think, precipitously lower than it was two years ago, five years ago. Now that's also because of that, competition has gone up a ton. So everything's getting more expensive.

It's going to get more expensive and potentially the flow of dollars you're competing. There's fewer. I guess that is just competition. You're also competing over all these adjacent things. So ad spend, you're competing over the same eyeballs on the Facebook's networks and all these things. There's all these sort of effects. I have not gone deep on analysis, but I've been trying to think about what is going to happen in the app ecosystem as this stuff plays out. And I think we've seen it. Yeah, there's just going to be more competition. It's going to be tougher. The dollars on the supply side, I don't think I've necessarily gone up. I don't even know if we would have a way to easily see that. We'll get numbers from app figures and stuff like this towards the end of the year, Sensor Tower, towards the end of the year on total consumer spend. But I'll just say from a naive analysis of our macro data, it has not

seen that spend has rocketed up in step with the amount of new apps we're seeing be created yet. This sort of comes back to where we started, which is I still think we need better tools to help people make better products. We brought the cost of building down, but we need to bring the quality up on average. This has been also why I keep asking people and I'm keen to see because even if the quality just stayed the same and the number went up, let's say we doubled the number of apps, quality stayed the same if you could measure quality by some abstract number. You would have twice as many outlier great apps you would think - Quality's been the same. Yeah, that assumes that quality stays the same, which basically assumes that there was bottleneck on engineering was the only bottleneck and that engineering or development costs drop, suddenly all this excess of taste or whatever can flow into the app store. There's certainly excess

initiative. I'm convincing myself of your point, Andrew, is that there's now plenty of initiative, but is there actual ability? Maybe to some degree, yes, but maybe not universally. But I think people can learn. Certainly. That's why X1 is cool because people can learn. I think everybody who's made a great app made a bunch of bad apps before that Usually. Yeah, Exactly. Exactly. It's rare that somebody is just born. I'm sure there are counter examples, but it takes a lot of years of using software and getting opinions and making bad apps.

Talking to users. You have to be able to iterate a ton. So whatever is built has to be set up to be iterable. Yeah. Which is the case. I mean, the cost of production has gone down, but so has the cost of iteration. All of Those Costs have gone down. Do you get signal from average ratings across the whole app store? It certainly exists. That is actually a very poor measure of quality, unfortunately, because so many apps are gaming it. I mean, this is actually something Apple just clamped down on in the last week. Which Would change the numbers drastically, Right? The fact that they're - Apps have been... I mean, I've had this question for a long time.

Why are all these apps who seem really crappy? Why are they 4.9 stars? In the app that I've poured my heart into for a decade is 4.5 stars. And the answer is when you ask in onboarding and you prompt write, you just get a volume of people who just don't even care and hit the five star and hit submit. So unfortunately the star ratings are so gamed at this point that it's a very poor quality signal. The one interesting thing, I think Ariel from Apfigures was doing this. He devised some ratios of the star rating of written reviews. What number of one-star written reviews do you have compared to the number of five-star ratings? Because it's easy to get somebody to tap the litle five-star thing and hit submit. It's a lot harder to get them to go actually say something good and meaningful about the product. That's game to some point too, where people are still buying reviews and things like that, but Apple's cracked down on that quite

a bit. So it's harder and harder to actually fake their written reviews. So that may be a better signal quality adjustment is to look at written reviews and do some ratios there. I'm glad to hear they've cracked down on it because at the mine company, we had some blatant copycat out of Eastern Europe that clearly just bought a ton of ratings and reviews and we collected a bunch of evidence and sent it to Apple and they just didn't do anything. But if they've cracked down on it now - Three years later, they cracked down on review farming. It's progress. Andrew, your answer, I think it's probably just total spend. You know what I mean?That's probably our best. Well, you got to probably per capita, probably divide that by humans on earth or something like this. I Mean, it's always retention though. Can you see retained spend, non-churn, churn, some indication of churn?

Yeah, that'll measure existing players. The last two years, the answer has been yes, which is there more software functionality that can be extracted from phones specifically, but just in general, personal usage of software. Is there more depth than we had pre-AI? And the answer is, well, absolutely yes. I don't know if it's 2X, but I feel like it's a huge increase in how and what people are doing with computers than they were two years ago. And that was just a factor for making computers smart and letting them talk, which is no small feat. But do we think we've hit bottom?

And that was like, okay, that now we're actually maxing out what humans can do with computers? It's like, probably not. There's probably more depth to it. Will we hit... LLMs were just such a step function across everything. They made almost everything better and faster in many different ways, both in what's in the apps, but also how they're made and all this other stuff. No, that's not the same level of innovation that we're going to have. Again, talking about a tasteful entrepreneur making something, that's a very small needle point in the total value of apps and humans using software as they attack a niche. Again, I'll go back to ATC, this app I've been really enjoying. It's a very well done app that in theory could have been built a while ago, but nobody did. And it's a fresh take. And I have to think that, and it's doing well in a sense that I see it everywhere. People are talking about it. It's become a thing and I

have to think that that's still possible. And even in chess, even in these long, even in Strava's category. I have to think that there are unique things to say still. For sure. Yeah, absolutely. But the chance of finding that this is so infinitezmal that going back to the venture thing, it's like that makes it a very, very hard venture bet. Well, it's also just hard to disrupt someone else's network effects. But I'll tell you two things I'm not worried about. I'm not worried about apps not mattering anymore because everything agents, not worried about that at all. And I'm not worried about everyone just building their own bespoke software for everything. But both of those things have come up as major threats or topics or whatever. And I just think there's no chance of that. One of the classic mistakes in a venture-backed business is you get some company, they're crushing it, they're super confident, they're growing really fast. And then the CEO's like,

"I don't like Salesforce. We're going to build our own CRM and it's going to be better." And then it becomes a total boondoggle. It's the same thing. Let someone who's totally focused on making a thing great make that thing and then use it. And maybe there's more of a personalization layer on top of that. And I've built software for myself and I think everyone will do that to some extent. And that might drive down the cost of buying it off the shelf, which is good, I think.

Yeah, it competes with the incumbents, the Actual commercial solution. It can't be outrageously expensive anymore. But people are still going to want to use products that other people create. And then secondly, the form factor might change, but people want beautiful visual experiences. It's not only going to be agents talking to each other that's going to be under the hood. But then people want to interface with something using their eyeballs. Yeah, Something that's tasteful. That's artistic at the end of the day. Somebody was talking about the Vision Pro and how if we were at software maximalism already, the Vision Pro had some really cool, I've never actually used one, so I should shut up, but had some really impressive demos.

They have this dinosaur wall demo. These are amazing experiences that are very short. They're very minimal. And if AI was truly replacing, you would have these insanely deep experiences on these niche platforms and it doesn't feel like we're there or even close to there or anybody's even working on it. The bottleneck does not seem to be just purely the ability. There's a lot of pipeline stuff that goes, we're not at visual rendering capabilities for asset creation on these things and stuff, but a lot of bottle bottlenecks still. And to be honest, at the beginning of Opus and the new model wave, I was like, oh, kind of like it's software.

And we've seen certainly parts of our software have become less compelling than they were pre. But yeah, I mean at the risk of saying like, "Oh, of course my business will never be disrupted," but I think I've come around to the same sort of belief. Well, on that high note, let's move on to the lightning round. There's three questions I ask every guest when I remember at the end of the podcast. By the Some definition every guest.

What's the most impactful experiment or change you've seen this year? Not operating apps yourself, maybe in a portfolio company or across the industry, what's the most impactful thing you've seen this year? Okay, I'll give two real quick. I will give one for myself, which is that I started building my own CRMs and productivity tools, but the impactful part is teaching them to learn and improve. That's what I use Zoe Computer for basically. I have a CRM, I have a to - do list, it's all connected to my email. I'm not an engineer, so I'm not going to blow myself up like I would using OpenClaw, but then I actually have it evaluate its own performance and get better on crafting priority to - dos for me. And that has been really, really cool. And I'm just excited to see what happens in software in general with those sorts of techniques. And then I have a portfolio company that built their own internal analytics tool that's just ridiculously powerful and anybody

can use with natural language. And so that's one thing on the portfolio side where I just see the democratization of analytical capabilities within companies super, super powerful and I think will lead to more velocity and better decisions. So those are two. This is a whole nother pod, so I won't fully chase this rabbit and go down this rabbit hole. But my biggest concern with the internal analytics stuff is, and I've heard from people who've had this issue, is data quality and relying on the answers you get out of your internal analytics pipeline is scary to me.

For sure. I think you have to put a lot of effort into making sure it's good. So it's one thing to set it up, it's another to improve it over time so that it actually understands your data structures and you don't have the garbage in garbage out problem. But if you can get there, which I do think is possible, then you do have a natural language analytics tool. That's just not having the bottleneck of data analysts and data scientists or asking full stack engineers to run SQL queries for something. I think that's a game changer for Companies. I've been using Amplitudes AI for my product analytics, Enrico for my RevenueCat analytics, and it's pretty dang Good. We got to get all these folks together. We should have a Big chat Room where we add all the agents and let them talk to each other.

All right. What's the worst experiment, the biggest fail that you've seen in the past year? I've seen a number of security related incidents, and I think this is just a byproduct of moving fast with AI. I think people need to be conscientious about what vendors they use as well and how are their vendors thinking about governance? Or is it just max velocity ship code? The industry is going to figure this out and I don't think the right answer certainly for a company is to be like, "Oh, we're going to be conservative and go slow because you're just going to get crushed by the competition." But I have seen multiple serious security incidents that I don't think would've happened pre-AI. And so I just think it's something for general cognizance and awareness and thoughtfulness around the trade-offs, both internally, but then also in terms of the tools you use and what are their principles?

Yeah, that's a really good one. Last question. Growth would be easier if... Yeah, okay. So we talked about this earlier, but I'd like inference costs coming down. Three tokens. That's a great one. That's a Great one. Super cheap inference that's really high quality I think is going to... I mean, it's just going to feel unbelievably magical. And I agree it does seem like we're kind of going the other way right now. I think probably the best models are like we can't just light money on fire forever, but I do think there's going to be more model routing. There's going to be more building of local models. There's going to be better physical infrastructure and all these sorts of things. It's going to come down. I don't know exactly when or by how much, but I think that's going to unlock an enormous amount of growth.

Yeah, such a great answer. Thank you so much for joining. This is super nerdy but super fun. I don't think we've ever gone this deep on venture capital before even explaining what now we is. We had Andrew on Twice. We did a 101, we did 102. You know what I mean? So this is great. I think people take a lot of value away. So thank you Andrew for joining us. Yeah, thanks for having me. Good to see you guys. Thanks so much for listening. If you have a minute, please leave a review in your favorite podcast player. You can also stop by chat.subclub.com to join our private community.