5 Solo Founders Called In With Their Hardest Problem | Solo Founders Office Hours

0:00Really excited to do this. This is a new experiment. Ari, I think that you're going to kick things off here. Yeah, sounds good. I'm really stoked. So first up, we have Utkarsh. He is an ODF alum. Right now, he is working on an open source app that allows users to play PC games on Android. He's got a million users, 50 ,000 Discord users, 10 ,000 GitHub stars, all organic. Utkarsh, take it away. Thank you. Hey, Julian. Hey man, good to hear you and see you. And it's been a little bit since ODF. I've just always been so blown away by what you've been able to do on sort of the community side of things with what you're building and the things you're doing with open source.

0:45So let's get into it. What are you trying to work through right now? What are you trying to figure out? Yeah, firstly, thanks. It means a lot to hear that. But yeah, as Ari mentioned, working on GameNative, it's been growing well. and it's getting to the point where you know it's it's it's a really big product a lot of people like 1.25 million people are using it now in every country in the world but I'm trying to take it from like this cool open source product which people find very useful they use it every day into something that is more of a business and you know that that can sustain me and potentially our employees.

1:24So, you know, I've been trying a bunch of things right now. Most of our revenue comes from selling premium user features. So that's things like a role on Discord or early access to builds, but, you know, playing with some other ideas as well. For instance, we started recommending and selling games via affiliate links with a partner store within the app. so that's kind of an experiment that's going on uh like we get a very small percentage of each sale so that's not a huge money generator uh we're also doing now sponsorships within the app yesterday we had our first game sponsorship within the app where people will see like a sponsored little icon on the top left when they when they open up the app and they can click to wishlist that game and we get paid per wishlist cool so what are you trying to figure out right now like just monetization or something else yeah yeah my main my main issue here is monetization like making this into like a business got it uh well i'm really curious you know you said a little bit about kind of what you're charging users for right now can you say a little bit more about that just really briefly yeah um basically it's uh it's it's like a um 10 or five or 10 uh dollar monthly subscription for that you will get a role on discord you get priority support getting games running um you get uh early access to builds of the app um and that's pretty much it uh and so we have we have a few hundred people who do that right now uh which gives us some money but obviously it's not it's not a huge sum yeah yeah for sure i mean i'm really curious you know what you can do on the front of building things that are just more impactful than what you have on open source uh right now that that will essentially allow people to have really good reason to pay for you because all those things sound great but those things all sound like sort of bonuses right versus core stuff and i think that you know the reality is that people are really excited to pay for things that they find valuable right now it almost feels like a patreon type thing you know where uh where it's sort of like a fan club people who just want to support you uh people are doing it sort of not because it actually makes a ton of monetary sense, but more because they want to sort of, you know, show their support and get some like little additional benefits.

3:46Yeah, that's exactly what it is. So you think it makes more sense to charge the users rather than going for some of these business use cases that I was going for? I mean, it could be interesting. You know, I do think that gaming is, you know, has its own subtleties as an industry. I think the main thing, though, is like, what are people doing without you? what what are people doing without game native like what what does game native enable people to do that they just really can't do anywhere else like maybe maybe say that really briefly just for other people who are listening in and hopefully people are weighing in with other thoughts on the on the chat as well uh because this isn't something that you know i'm i'm trying to be like the definitive voice here i want this to be a thing where other people are having discussions what what would you say though just for people who are trying to familiarize what would would um would somebody be doing without GameNative?

4:35Or what does GameNative truly unlock for them? I mean, what it unlocks is being able to really play their PC game libraries on the go away from their home PC or without a PC at all, right? Nothing else even comes remotely close, right? All of the other things that you and I have talked about that are trying to build something around this have nothing on GameNative. Well, debatable. there's a product called Gamehub which I obviously think is inferior but some of our users compare the two and say that they're pretty neck and neck so what we do uniquely over Gamehub is that we are open source we are transparent and people really care about that they really dislike how proprietary Gamehub is they don't trust their data practices all of that kind of stuff so it's not just that we let them do whatever Gamehub does plus more we also do it in a transparent way we have a lot of user trust and that is really part of our secret sauce as well that's great I think there's a way to monetize while also maintaining user trust I think that I think that user trust doesn't matter if you don't exist yeah true so it's better to find a way that and honestly like I think the other thing about trust is when you're open with people, people will step up and understand.

6:03Not everyone, right? Obviously, there's this saying that sometimes the people who pay the least are the ones who complain the most. And this is more for B2B stuff, but essentially when you look at a lot of customers, the ones who have the most customer support requests are the ones who are paying the least in terms of SaaS spend. so there is something to be said about you'll probably get some people who are upset but I think that you'll also get a lot of people who your honesty and transparency about decisions that you make will be something that they understand that they end up supporting. Makes sense.

6:44Yeah, I mean I think that's good homework for me go back and see what part of the experience I can see that users could be charged for. I also think that it's not even about reversing something that you've already done. I think it's about what you do in the future. It's not about taking things away that people are used to paying for. It's about creating new things that they haven't paid for yet or they haven't used yet. The other thing is grandfathering people, all that. You have great sensibilities here about what to do for people. So I'm not worried about you doing wrong by your users. I know that you'll do right by them.

7:21Fingers crossed. Cool. Awesome. That was amazing. It's really good to hear from you. Congrats on 1.25. Thank you. Thank you. It keeps growing. It keeps growing. Now, give them a way to actually pay for something that they'll really want to pay for. Because 1.25 and only a couple hundred people paying you, it shows that you haven't yet offered something that's actually worth paying for. And I suspect that once you give that opportunity, people will take you up on it. Understood. Thank you. That's really great feedback. Thanks for being part of this and for sharing this so openly with everybody.

7:59Ari, who are we going to talk to next? Awesome. Next up, we got Ryan, the founder of Crash Labs, where he's building benchmarks and RL environments from real professional work, which he then sells to Frontier Labs. He raised a pre-seed, previously scaled an AI support code to 84K ARR, and before this, he was doing data evals at Cohere. Ryan take it away. Hey Julian thanks for having me on. Of course yeah this is great. Yeah so uh just a bit of context on where I'm at right now uh I actually went through a co-founder breakup as of like last week so I've pretty much just been kind of working these past two weeks trying to figure out what kind of direction to go in and also like really just building out like an initial version of uh an agent benchmark which I called agent mafia which is going to be released right after this so uh yeah exciting i've just been like spending the past few days putting my heads down just really trying to um build different things that i think might be useful on the benchmarking side and seeing what results come out of that cool what are you trying to figure out right now or what do you want to talk about yeah i think one of the things i've been like really wondering especially now that it's been like almost exactly a week uh since i officially became a solo founder is like like how to you know adjust from you know having someone to consistently spin ideas with or consistently work with to like now kind of being on your own and doing stuff so everybody's different uh first off um was the co-founder split was that something that sort of you you wanted or was that something that kind of you know kind of you feel like was an unfortunate thing or or you know what i mean like not yeah it was an unfortunate thing negatively or anything like that of course but i'm just trying to understand like the dynamic of it um and i hope you're i hope you're doing okay yeah yeah no thank you i'm doing well um yeah it was like uh it was more like an unfortunate thing i think it was just like due to a lot of personal reasons that we couldn't really control and like also like we still are like really good friends and everything but um it just like was the startup itself that was taking a toll so um yeah we decided to split up and i think like we we've just been i've just been working on like how do i kind of adjust to being a solo founder because like the entire time i've had a startup at least or been running a startup i've never really like been on my own before so just figuring that part out has been like something i've just been trying to work through the past few weeks what i will say is that um being a solo founder is not easy like it is really hard uh in a lot of ways um it's particularly it's hard in the zero to one phase before you have customers before you have a team once you have customers it's this incredibly orienting force right um once you have that you really know like what you're doing who you're doing it for and of course your business will change you might sort of throw away the product and build a different product in fact a lot of solo founders in sfp they've actually started with a bunch of different you know ideas and sort of what we call pulling the thread so i i will i where they kind of go from one thing to the next thing but they're all sort of connected it's all one thread what i will say is that um that you should be be easy on yourself about the fact that you will have you know a tough time sometimes balancing should i be talking to customers should i be building um the answer is almost always you should probably be spending more time talking to customers or or users if you have them and then building is is sort of like the very easy thing to I think you said you're an engineer you have a technical background right well well even if you're not an engineer nowadays I think that building is like the addictive default right because it always feels like you're making a little bit of progress but really the rubber only meets the road if you have you know paying customers or you know if you're not doing something that's that's paying customers you have some scale of user base that you're focused on in order to monetize, you know, the larger consumer sort of product.

12:21I would say that the main thing though is if you can just be easy on yourself in that regard and then really try to make sure that you are like being honest with yourself about what you're actually doing on a day-to-day basis. Because I think that some people find it very clarifying to have somebody else with them and sort of like a group fitness sort of group workout thing you're more likely you know if you don't already have a workout habit you're more likely to go and work out if you've made a commitment to work out with someone every day or you know every number of days so I think there is this aspect of when you're solo you don't have an accountability buddy and I think that that really makes a big difference especially in the early days when it's so nebulous right if you have customers you're like i want to grow you know x percent week over week or something that's quite clarifying and again you could have this own sort of measure of success but in the early days you just want to make sure that you're spending your time effectively i'm curious right now where you think you're at in terms of you know i know you're you're building something here but like where you're at in terms of like actually having people touch it um like getting in front of people how do you think you should be spending your time how do you think you are spending your time um yeah i'm just curious to hear a little bit more on that yeah yeah so the rl environment space is very interesting since it's also kind of a very new space and there aren't exactly many uh like you know how two guys when it comes to um you know working with researchers and doing stuff like that and it ends up being much more of a service right now as opposed to just like oh I built this data here's the data like it's a lot more about like curation and working with researchers and actually figuring out what are like the deficiencies that your model has that we can provide and so one of the things that I kind of wanted to do really fast was just like how quickly can I just put something out there because what a lot of when you don't have like inside connections to you know research labs a really great way that I learned is to get inbound from other you know people working in this space is just like building benchmarks posting them publicly and then getting um like just have like people whose malls aren't using posting uh performing the best on the benchmarks usually tend to kind of inbound uh and be like hey um where is this data set like or where's this rl environment could i like perhaps train on it and so that was like i think task zero that i wanted to do was really just how can i get something out there as quickly as possible and just gain momentum both for like the product itself and also for myself as well like i didn't want to spend too much time like you know worrying about oh is this a good idea or not i really just want to like build and just put something out there so the benchmark itself is a a bunch of frontier models playing the game mafia with one another and it doesn't just decide on like who wins but it's also measuring like what were their deduction deduction abilities like who was the best liar who lied the most frequently and who was able to deduce things that weren't you know inherently obvious uh at the at that point in the game and so it tests for a bunch of social reasoning and multi-agent communication work and i think my goal is that i want to be building aural environments for continual learning which is this branch of ai where it's about how can agents consistently recursively improve on new information once it still has retention of its old information.

15:55And so right now, I just kind of want to build in that direction. And this is like step zero into getting there and showing people, A, I have research taste, and B, I can build good benchmarks. I think it's great. I mean, I think that this is the exact right move is to go out there, put something into the world that people can react to. If you don't put anything into the world, you have to be sort of quietly sort of pinging people and doing outbound. but in your world what you're working on much better to put things that cause people to come to you what's your what's the what's the response been like when you actually get into these conversations uh with people after they've reached out um so far i haven't published the benchmark yet it's going to be published later today but um what i this is a few researchers that like i kind of have spoken with back and forth is that it ends up being much more of a like curated data set service rather than just like oh i have something let me sell it to all the frontier malls at once and so it ends up being much more of almost like consulting in a way where it's like we are going up to these people and figuring out what are like their unmet data needs and how can we kind of get there and it's always changing as well because the space is moving so fast and there's already so many vendors who are focusing on different stuff so what i want to do is concretely a few things one is just like figure out what is like the specific niche domain of rl environment that i can build that is unique to me that I can service to these frontier labs and kind of starting off with just agent mafia which doesn't seem inherently like as useful as like enterprise data for example but it is something that I have a lot of interest in and I have I do play like a lot of social games so it was easy for me to comparatively build the verifiers for this rather than just like any old enterprise tool essentially and so I kind of just wanted to really dick my toes into something that I was uniquely suited for and then seeing what the response was.

17:47And I kind of used that feedback to, you know, start more conversations with customers or get in front of people. So that's the approach right now. That's great. You know, one thing that I've seen work pretty well is, you know, this idea of highlighting sort of the gaps, you know, that that a company or a lab might have so that they can say yeah we do have those gaps i can see that uh in the benchmark um and you're now proposing that you could actually help us you know fill those gaps and and solve for a like an uh an identified and agreed upon problem versus you know as you're saying just selling generic uh generic stuff one size fits all to everyone um yeah it does look more like a service but it might be a really valuable one.

18:34Exactly. Yeah. I think it's like pretty much the go to market strategy in my head right now is like show them the problem and then sell them the solution essentially. And I think that's actually, even though that might sound counterintuitive, that's actually the way a lot of frontier lab data vendors have been working right now. It's just like, you know, like showing that, hey, this model is not super great at design. And then a few months later, having the research lead say, hey, can we train on your data? So that's kind of just the approach I'm going for right now. Sometimes when you build a new technology or build a new product, this is less related to you but more of a general idea for other people who are listening or watching.

19:16When you build something, sometimes it's something that people haven't even considered yet. you know there there's a there's a an sfp founder who's building this really incredible product um and the user behavior is just something that nobody would have uh would have considered before because it just didn't exist but people are addicted to using it they think it's such such a great thing it's it's like showing them how much um some of the things they own are worth and like the usage and the sort of like repeat usage is really high because it's just kind of, it's tapping into something that is a part of human nature and the human interest, but in a way that's never been done before.

19:59So I think there's a lot to be said about you could help people understand sort of their problems or their opportunities. And then if people feel like you have a credible job, you could do a credible job of actually helping them do that, that I think you've got a really good potential business. Yeah, yeah, absolutely. One kind of small question I had is like, how do I communicate, I guess, like the, like I think one little fear I have is that like, since I'm a solo founder now, it might be harder for labs to take me seriously or for even labs, like once they peek under the hood to think, oh, this guy is just like a solo person instead of like an established team, for example.

20:39So having that bias of like, oh, will he actually be able to provide us the service we need if he's just like a single person uh so that's like just like a fear i do have now that i am a solo founder yeah i mean well it's it's a it's a valid fear to some extent um if one people don't actually believe that their problem is significant and two you have to be competing head-to-head with lots of people right um so hopefully you figure out a way to build something and acknowledge issues or gaps that other people are not necessarily helping them solve for because if you're if you have head-to-head competition i mean like i think that there's like the like the peter teal like competition is for losers yeah and i think there's a lot of truth to that it's like and i another way of thinking about it is uh this uh this author kevin kelly uh probably multiple decades ago um wrote this this idea you don't want to be the best you want to be the only and uh so i think it's more about less less about like oh what what does the team look like and more like oh wow is this like is this like not just the best but it's actually the only uh person who's going to be helping us with this um you know i i think also from from what ari said i believe you have a like credible background um it's not like it's not like you're just some random yahoo uh you know you've actually been you know working in this field like somebody should in theory consider hiring you you know to join their team or to you know to work on sort of a consulting basis, you can use your background as a way to make people say, I don't even know if this is a product.

22:17I don't know if it needs to be a product. I just think that I should be talking to this person based off of proof of what they've done in the past and also proof of what they're putting out there into the world now, right? This thing that you're about to launch, it sounds like it will help people understand you better and be excited about what you're working on and want to talk to you because of that as well. I think that a lot of this stuff around being solo is, is probably more in your mind than it's going to be in others. I think there's a pretty easy answer to it, which is, you know, I'm, I'm starting out, I'm just working with a handful of people, right?

22:52I'm not, I'm not like trying to scale myself significantly and I'm going to be hiring people and bringing on more people. And we're going to have obviously a lot of leverage with AI as well. Yeah. Yeah, absolutely. That's a really great way to put it. Thank you, Julian. Very nice to chat. Alright, Ari, who's next? Alright, we got Charlie Kahn, founder of Brio. He's building wholesale ordering in plain language. Say what you want, POs hit the ERP in 10 seconds. First customers moving 15 million GMV off of Shopify. Second design partner recently asked to invest. Before this, he bootstrapped a GTM think tank from$900 to 3 million ARR before exiting.

23:33Charlie, stage is yours. Like Ari mentioned, so I'm building a wholesale operating system. So buyers and sellers basically just say what they want in plain language. Then the order builds and prices and lands in the ERP without anyone kind of touching a back office system. Super early stage. We have two pilot partners that we're talking to almost every week. And then I'm speaking pretty regularly with other customers who are on our wait list. And that AI layer is basically the story that's gotten me every call with a supplier. But once I'm on those calls, what happens is they pretty much immediately start asking for features.

24:06And I can go into some examples. But my question is really about sequencing and kind of what to focus on right now. And whether to pivot to the workflow features customers are asking for now or go straight at the much bigger AI layer they're not asking for. But that actually sort of makes those workflow problems disappear in the first place. Why do you think they're asking for the features? Yeah, it's a good question. I think it has a lot to do with the problem. Everyone sort of agrees the problem is the same shape. Buyers and sellers know exactly what they want, but it takes them weeks and weeks to transact.

24:40But what's interesting is that the customer thinks in the world that they know, and that's pretty much features and screens and tabs and upload buttons and download buttons. And there really isn't anything like our product at all, and so they can't imagine it. And so they're trying to solve in the shape that they know how to solve because what we're doing really until a couple years ago didn't exist. You know, it's a really interesting point also. It reminds me a little bit of there's a company that I worked with and worked with before starting Solo Founders. It was sort of like one of the main inspirations for starting Solo Founders program, a company called Magic School.

25:21and the founder, Adil, he was a former high school principal and he wanted to get teachers to use ChatGPT. He wasn't trying to build a business. He thought ChatGPT was transformative for teachers who were really being stretched in terms of their bandwidth. And so Adil tried to get people to use ChatGPT. He did a big session for all of the teachers at his school and no one ended up really using it in any sort of meaningful way. Sure, they used it like Google, but they didn't use it for the tasks that he felt were really important. And as it turned out, he felt that the thing that was holding them back wasn't the fact that the pool wasn't, like the underlying technology of the AI wasn't powerful enough.

26:06It was the interface. And that when you had this sort of black box, this sort of empty sort of chat window, it was really intimidating because it could both do anything, but by the ability to do anything it could also kind of feel like it couldn't do anything at all you'd have to really sort of like go through this um this cognitive friction of trying to figure out like what am i going to put what i'm going to prompt right um i do think that in a lot of ways what people tend to do with the with the sort of ai solutions is they tend to move towards something that's like kind of an open-ended sort of thing like you know chat gpt or you know claude bot something like that but what a deal did maybe this is useful for you maybe not um is he he essentially was like no no like people like tools that do things they like buttons they like um they like something that's not like completely open-ended because that's what they're used to so he built what i would describe as like micro apps or applets um inside of magic school which is essentially there's now i don't know probably a hundred of these tools in magic school uh but when he started out i think there were three or four maybe five and the idea was each tool had a very specific purpose now somebody who's who's critical and trying to poke holes would say oh they're just like these little wrappers on top of gpt right but the reality was that the little rappers were the difference between somebody not using chat gpt and somebody becoming you know a multi you know hourly active user or whatever of magic school the company went from 100 users the first week to a thousand users the second you know now it's used by millions of teachers all over the world uh the last i heard one in four kids in america uh is at a school that has teachers using magic school the um the reason i say all of this is because i don't know if there's actually um if it's actually as clean cut as do the ai thing or do like the the feature thing like there might be something in between which is sort of like these micro apps um the other the other thing that's that's interesting is like um what it sounds like they're telling when they don't have feature requests is they're actually not asking for features they're asking for jobs to be done right yeah no that's exactly it yeah so from that perspective it's like the micro apps uh that magic school had which i think is really brilliant model for them and maybe other companies as well is the idea that each micro app is like its own job to be done and by constraining that and then letting people choose the job that they want it's sort of the best of both worlds the flexibility of an LLM while all, you know, in a sort of a chat-esque feature, while also the specificity of helping people really narrow in on the thing that they actually need doing with sort of the UI that corresponds to it.

29:08Yeah, I think that's an interesting way to look at it and kind of like mirrors some of like the thinking I've done too about like how do we like figure out what step to go through first? Do we kind of start with like version A that is like maybe more of a basic portal that that has the buttons and features and like add some automations along the way or version B, which is this bigger brain, the stochastic router that can sort of like intercept intent and figure out how to route it and deliver these deterministic outputs, which is really like key to wholesale. Like it can't, it can't, if there's 98 queries happening, it can't be right on 97 of them.

29:41It has to be right on, you know, all 98 every single time. And that's the challenge. And so it's like, it's a pretty big problem to solve. And so I guess like my real question underneath this is like, Like the incumbents are all playing this game of like adding features and tools. And eventually they're going to add AI tools. Like one of our customers is recording their calls with sort of the, like the leader in the space, ElasticSuite. And they're taking that, like they're sending us some of this information and they're asking about AI and they're kind of like, well, we're going to add a tagging feature or things like that, kind of like what you're saying.

30:14And so what I don't want to do is kind of get caught up in this race and basically just be another portal with some automation or buttons when we could be this entire brain. Because today, a buyer and a seller are both really good about saying what they want. They can describe it really clearly. And they're very frustrated that, you know, for example, it takes two or three weeks of back and forth, you know, email and CSVs to build an order. And yeah, I'm trying to figure out how to focus time right now because our big product actually eliminates all of these little workflows and sort of features entirely.

30:46the only the only challenge about the big product is whether or not people actually use it um and then the other then the other big challenge is like or the question that is like do they need to use it or can or can something like an ai use it for them the um the the thing i will say though is um it sounds like people want time to be compressed um so the big question then becomes why isn't time being compressed do they actually want time to be compressed or is the is the revealed preference actually actually different um is there some incentive that actually causes the time to you know to draw out the way it has um i don't have that answer um and and and it might not be apparent uh at the surface but sometimes things that actually take long um take long for a reason.

31:39It's just not the reason that everybody believes they do. So that's worth considering and sort of looking into maybe further. The other piece though is, I mean, like if you could just figure out sort of the sort of problem that is highest rate of occurrence, like highest frequency of problem, and then also balance that with sort of severity of the problem um i think that that could also inform how you think about this and what to prioritize okay yeah i mean on your first point it's super interesting for like why does this exist in the first place um and i've spent a lot of time thinking about that it's it's like pretty much the the core thesis of our our product which is that for 200 years buyers and sellers have communicated in natural language like when you want to buy something even today an order portal exists, you pick up the phone and you text Joel.

32:31And so there's been this like historic trend of natural language. It's always been natural language. And we actually have like the opposite problem of consumer commerce, where now we're trying to move to natural language from like click and drag like GUIs, like Shopify. And so we built Brio. The idea is like, as I've been thinking and iterating, it's basically just like texting Joel. It's just transaction makes the next one faster and smarter. And so that's our thinking on like why people will use it because we're not asking them to change behaviorally how they've already been ordering. We're just sort of standing in the middle of it.

33:06But again, like I said, that's a much bigger problem to solve than the workflow challenges. Yeah. It's interesting too because, you know, there's also this interesting aspect of does anything need to change or can you just be ingesting um in the in the near term um and and essentially like have some of this stuff become more emergent after you're just sort of taking in information do you need to do you need to try and insert yourself in a way that is like highly visible early or can you slowly start to emerge um and add value like throughout parts of this that kind of like help compress the the time cycle it's very interesting this is really cool um congrats on the progress so far convincing somebody to move like that much GMV onto what you're building is a great sort of sign that people have a real problem and they're willing to try something new.

34:00Yeah, thanks so much for your time, Julian, and appreciate all the feedback. Amazing, awesome. Ari. Andy's our last founder for this session. He's building Inphenomni, where he's essentially trying to build Prompt to Robot. Describe a robot, get a design, plus 3D printable kit. He's very early with that. Before he was working on this, he was a software engineer at Zoox and prior to that, Oculus. He even shipped an iOS game on the side that got some press. Andy, it's all yours. Thanks, Ari. Hey, Julian. Hey, Andy. I'm really excited about building a robot. Oh, yeah. Me too. And everyone, I've shown the early builds an idea to.

34:42I've just been like, oh, so I can print my own character and like when i have to tell them they're still still on the way it's a bit like um more fuel for them to keep building so yeah uh i mean ari mentioned that you turn like a photo or drawing and it returns a snappable assembly kit uh so this bomb itself is software and hardware which is like a pretty big scope. So I'm trying to figure out how to tackle this type of problem because as a solo founder, focusing on a software product and then shipping that to customers seems like the loop there is faster to iterate on. Like software, I can serve digitally within a couple minutes, the hour based on kind of feedback that I've received from someone or a company.

35:42Whereas this like hardware kind of product is like the end product is like this, the kit itself. But to go through that iteration cycle is a little bit longer and a little bit more, has more stages for things to go wrong. and so kind of like I think my question here is like if I had an extra pair of hands do you think it makes sense to be even thinking about delegating maybe like splitting the software and hardware problem to give like me time to focus on software on some days and they focus on hardware and then we can like switch back and forth to keep everyone in the details or have, I guess, you worked with any other solo founders who have tackled a scope this ambitious?

36:37I'm happy to talk about your specific case, but I'll also just level up here for a second and say that this is like every company, right? When you're starting solo, every company has a bunch of things going on and you have to figure out how do you split your time between those things. In your case, one of the, or two of the many things going on are software versus hardware. I think there's a couple of questions I want to ask, though, before I sort of give some thoughts. The first question is, if you were to, like, snap your fingers and you had, like, an amazing person who could go full-time with you and you had, like, money to be able to afford, I don't know, whatever costs you have around building this.

37:23like I'm not going to give a specific number but just assume it's whatever money you would need to do this like what would you be doing differently than you're doing today like how much faster would you be going like would you change sort of the the way that you're actually rolling out the product what would you be doing if you had like a person another person and uh and whatever money was necessary what would happen would I would invest all that into more 3d printing and hardware output on that front getting a library component. So putting together these companions, I've been trying out all of these different boards, connectors to like actuators and wiring.

38:05So just to really clarify, what you're saying is you would be testing a lot of stuff. Yeah. It would be less about commercialization. It would still be more about testing. It would be increasing your rate of testing, pulling testing forward. So you're able to make more progress on sort of understanding the various sort of opportunities with the hardware sooner. Yeah, but immediately to follow up on that was that because customers each kind of have been mentioning like their own favorite character as like a robot companion they would like to have on their desk, we would build those. It's not a really scalable approach, but we would build those in the meantime, get feedback from them in terms of like what they liked or dislike about the product, whether that's design or capability, how it's moving, and then incorporate those back into the system itself that's kind of the more long-term scalable approach.

39:07I got you. I got you. I mean, I also wonder if you really need to start with the hardware out of the gate or if you can do something where you effectively have like a brain transplant, you know, from something to the actual like physical object. So maybe it's like, again, this is just me, you know, spitballing here. So I'm not saying it's actually the right idea. But I mean, like you could just like have your phone run the character. It could be on a dock on your desk. um you know the character can't move uh like physically is it's just it's inside of like you know a frame on the phone but like now you've got the character you've got sort of like the thing and then you know eventually and that sort of like leads you to the next phase which is you know now you have like an actual like robot um and like when the robot turns on for the first time is now like freeing the the character that you've created from the phone you know and and bringing it into the physical world i don't know um have you have you considered something like that as a way to both narrow the focus early on and then also move faster yeah uh yeah so yeah i really like that idea um there was uh one of the i was considering yeah running some of the model compute it on the phone because everyone has a phone already um and some of them have gone if you have a more recent phone it's probably better at running like an edge smaller edge model and yeah and then connecting that like just kind of like as the shell to the the body until that part until the physical is ready um yeah has been an idea but yeah i think i think there's like a few directions i had to have to narrow it down to something like that just choosing to make it more or like in iterations rather than going for like a all the way to the moon type of approach i mean look like you could totally go to the moon um like directly like you know there's sort of that have you ever seen that meme of like how to draw an owl and it's like three steps it's like draw a circle draw another circle and then draw the right it's like it's like well maybe maybe you just need to draw the owl i don't know um like you know we're just meeting we're just talking about this for the first time um but you were asking about how do i like how do i make more progress faster how do i split my time and maybe the answer is maybe you don't start out by splitting your time maybe you make it so that people have these really compelling characters they're on their phone you have a you they dock on their phone um you know it actually compels them to not be scrolling on their phone when they're at their desk because now they need to have their character open um again like this is just me hypothesizing i'm not saying this is the actual right thing there's probably dozens of things you could do um but there's this aspect of like the only reason i'm going in this direction is because you were saying how do i narrow my focus or how do i like you know should i potentially bring another person on and like the answer is like maybe you should bring another person on maybe you should raise some money maybe you should you know build the the robot straight away um but i think that like what you're talking about feels like the robot like the movement of a robot sort of the physicality of a robot is really only part of what actually matters here and maybe it's even a smaller part uh of what matters because if you think about it um you know like you know not not not to be like morbid but like think about like really terrible sort of diseases right in humans like dementia for instance right the reason that dementia is such a painful uh horrible terrible debilitating disease is because the thing that makes a person a person isn't just their sort of exoskeleton right it is who they are it's their memories it's the way that they communicate um it's the way that they look at you and they recognize you when they see you right um so like the reason that it's such a terrible terrible disease and hopefully one that we can you know we can solve in the in the coming decade um is also the reason why um starting out less focused on physicality and more focused on these other aspects of like what makes a thing like a like an a a a being a being it doesn't need to be a human being but it could be a you know an ai being it feels like that's actually one of the core aspects of what you're what you're building and i don't necessarily know if the body is as important in the early days as as the mind um does that make sense yeah yeah that's a really good point and something, yeah, definitely I should reconsider.

44:04I think I also, if I may also kind of try and explain why I've been considering physicality. Yeah, of course. I mean, like, I'm not doubting physicality, but go on. Oh, no, no, I'm just, like, trying to have a, no, I understand. I just want to have, like, a, yeah, just, like, share some thinking and get feedback on it. Because it's just been kind of, like, me in the details and just kind of, yeah, working this out. So, yeah, the second half is, like, just like different characters from like a different cartoon show, like a Pikachu versus like a Snoopy versus like a Rover. They're going to all move differently in our, in the physical space.

44:40And so that's why I think I've added kind of like, kind of like an importance to that problem, solving that problem, solving like physical articulation. It's just kind of like if, if I brought Pikachu or I brought an owl into, into like robot form into our world, they, I would have to figure out, I would have to have a solution for the multiple to move in the way that they do differently, if that makes sense. Makes complete sense. And I think if you made owls, it would be a real hoot. Oh yeah. This is awesome. I mean, I'm really looking forward to seeing what you do with it. And I really appreciate you hopping on to sort of share what you're thinking through.

45:23And hopefully people who are watching or listening or we'll reach out if they have some relevant ideas or relevant connections or they just want to say hello and support because it sounds like a really great project. And I mean, look, your background's amazing for this. You've been working in robotics and software for a while at some great companies. So both consumer and, oh, I guess like Zoox is consumer too. It's just a little bit of different sort of consumer company. This is awesome. Well, thank you for doing this and for joining. everybody who came on, I mean, it's like, it's incredible to see the variety of things that solo founders are working on, the expertise that sort of is an insights that have driven these, these companies.