The Startup Ideas Podcast - If I wanted to build $1M+ AI startup in 2025, I'd do this

Starting point is 00:00:00 Cal AI is really cool. Like I downloaded that app and it's amazing. But there's going to be like a hundred of them in the app store in like two months. And then Cal AI will not be able to charge what they are. And I just go like those are good one-off bites. Those are good for going from a college dorm to being a millionaire. But they're not going to get you to a deca millionaire. I think like your equity value is just toast within a very short period of time.

Starting point is 00:00:25 Because I just think the pricing ultimately goes to zero. What do you see is like a big? business that you can implement AI or build an agent that can still be highly competitive and successful in two or three years. If you want to build an AI business, you need either a data advantage or a network effect advantage. It's never been easier to build a million dollar business and it's never been easier to lose all your equity value. So I just think that's a hard world to be a business person in. It's exciting. It's the most excited I've ever been for business because it's easier than ever to start a business.

Starting point is 00:01:00 But the worst businesses are the ones that are easy to start. Good to see you. Good to see you too, man. So today I was thinking we can talk about where are you seeing the opportunities right now with AI? Because in my mind, I know, I know you look up to the Charlie Mungers and Warren Buffett's of the world, value investing. You know, Warren Buffett and Charlie Munger, value investing, you know, what does that mean? It means like seeing a rough in the diamonds and sort of seeing opportunity. And I'm curious, where are you seeing opportunity in this world of AI agents and AI in general?

Starting point is 00:01:46 I think the hard part is that in a lot of instances, it's like a sustaining innovation, right? And what I mean by that is it's a new technology that will drop the cost. So let's say that you're, you know, we were just talking about recruiting businesses. So let's say, you know, you start a online recruiting business or you buy an online recruiting business. And they have a 20-person outbound email team. So, you know, you might think you're a genius because you're like, oh, I'll get a, you know, an outbound recruiter agent that will just replace all those people. And now my margins will be way higher.

Starting point is 00:02:21 But what actually happens is maybe for six months you get that arbitrage. But all of your competitors have the same realization within six, 12 months. and then the price drops and you're in the same business and maybe it's even more competitive than before. So I think it's one of the most interesting times to be an investor and entrepreneur because we've always been building sandcastles on the beach in tech. Like I think we all knew, you know, anyone can compete. You know, you start a to do less software business and anyone at any time can launch a competitor.

Starting point is 00:02:56 It's just that before there was a finite number of coders. the world. There was only so much venture capital money out there and there was just less competition. I think now you get far more competition in far more niches. And I think moats matter more than ever, you know, where, you know, a moat before might have been, you know, I make funeral home management software. And the moat is that it's too small of a market for venture capital to play in. And none of the other people that own funeral homes know how to code. But that's not a mode anymore. What a real moat would be is I have software that allows funeral homes to keep track of all

Starting point is 00:03:40 their inventory and trade it back and forth or buy it. And then you have a network effect. Now, someone can't just go vibe code that in a weekend because now you have a network effect. So, you know, I think social networks are really interesting. I think communities are really interesting. But I think most tools have become a hundred times harder. Yeah, I mean, I totally agree. I think I just want to get more tactical. I'm trying to think like what, okay, I think everyone agrees, honestly. I think everyone agrees that, I mean, I can't tell you how many distribution is the new moat tweet I've seen in the last like three weeks. It's like everyone has finally agreed that distribution is the new moat, that anyone could create anything. The latest thing is by the way that vibe coding is really, don't vibe. code your way to production code.

Starting point is 00:04:32 That's like the latest meme that's going around because I don't know if you saw there was this like viral post that this guy was using cursor and he ended up posting it production code and then he got hacked and bad things could happen. So I agree with all those things and I think 99% of people listening to this agree.

Starting point is 00:04:52 So what's our next step? Like I actually want to treat this as like you know and we've done this before. You know, you've called me. I've called you and called you and been like, okay, so what do we do here? So I've got, I mean, I've got a bunch of ideas. I wrote down a bunch of ideas for things people could build as we usually do.

Starting point is 00:05:09 The problem, you know, as I look at it, is almost all of them can be competed away really, really quickly. So I think it's been, it's never been easier to build a million dollar business and it's never been easier to lose all your equity value, right? So I think it's going to be really easy to have gold rush, you know, oh, hey, like Peter Levels can launch this flight simulator. make a million dollars, but is that a sustainable business in two years when there's a thousand of these flight simulators? To be honest, I don't know what the opportunity is.

Starting point is 00:05:44 I think it's a bit of a sad answer. Like I'm not seeing a lot of opportunities to go and buy or start businesses where I think there's a sustainable competitive advantage. I think there's a cost advantage, there's a time advantage. But I'm not, I mean, I would be really, curious to put that back to you. Like, what do you see as like a business that you can implement AI or build an agent that can still be highly competitive and successful in two or three years? Because I'm increasingly of the mind that for all tools, basically all middlemen businesses are gone. And I believe fundamentally that we are all going to go to the search bar and the new search bar is chat GPT. And you're going to query it. And chat chit will just build

Starting point is 00:06:29 whatever you need or it'll hook into whatever APIs you need. And so the only things that really matter at this point now are, you know, real world goods and services compute or network effects of some kind. I don't like that answer. What have you got? Like I'd love to hear your answer to this. I mean, my take is if you want to build an AI business, you need either a data advantage or a network affect advantage. So I think what does that mean for us? It means that we probably shouldn't be buying or incubating AI businesses today. So quick break in the pod to tell you a little bit about startup empire. So startup empire is my private membership where it's a bunch of people like me, like you, who want to build out their startup ideas. Now, they're looking for content to help

Starting point is 00:07:28 accelerate that. They're looking for potential co-founders. They're looking for tutorials from people like me to come in and tell them, how do you do email marketing? How do you build an audience? How do you go viral on Twitter? All these different things. That's exactly what startup empire is. And it's for people who want to start a startup but are looking for ideas, or it's for people who have a startup, but just they're not seeing the traction that they need. So you can check out the link to startup empire.co in the description. I think what I'm interested in doing, frankly,

Starting point is 00:08:06 and I'm giving my playbook here, but it's basically like I would either want to buy media businesses or incubate media businesses. Why? because I think that if you assume that you can vibe code software and you can create the most beautiful to-do list app, I saw someone vibe code their way to Things 3, a clone of Things 3,

Starting point is 00:08:29 which is something that I've used for many years now in minutes, and it actually works and it's beautiful. So if you can assume that, okay, you can vibe code Things 3, which for me was like an aha moment around like beautiful software, to do list app, then, okay, the opportunities in either buying media businesses, incubating media businesses, and then once you've built that network effect, then you can start saying, like, okay, now how do I get these people's data? That's your next step. Your first step is, how do I get the attention? Next step is how do we get your data? And then the third step is,

Starting point is 00:09:07 okay, how can I build assistance, co-pilots, AI software to actually monetize. and build a sustainable business. So what would be the sort of media business that you would want to buy? Maybe I have a different take, but I want to hear what you're imagining. Let's say you could press a button and buy, magically buy the company of your choice.

Starting point is 00:09:28 What company would it be? Well, I'll tell you, like, my framework for thinking about what would be a good company to buy. So I wanted to be a high value niche. Like, I wouldn't want it to be, you know, I don't, and I envy Mr. Beeson a lot of ways. but on on on the other hand I don't envy Mr. Bees like he's got

Starting point is 00:09:47 billions of people looking at him but they're low value I'm just talking purely economical like I want to high value he's got to sell stuff to kids and kids don't have money exactly but I'll give one I'll give two

Starting point is 00:10:02 businesses that I think would be interesting to buy that are high value that are probably under monetized one is tech crunch which just has a great brand name To you and I, like, we have so many memories from, you know, the Silicon Valley early Silicon Valley days, like launching on TechCrunch, that sort of thing, where I think it was bought by AOL, right? Yeah, I think so. Michael Arrington was the guy who started it and then he sold in like, what, 2008 or something?

Starting point is 00:10:36 Exactly. So I bet you there's a deal to be had where they're probably struggling. They're trying to probably figure out who they are. are what they are. But they've got a high value niche with a great brand. So I would buy something like TechCrunch. The other kind of, the other interesting thing to buy would be a very sought after event series. So my dream, you ask who my dream company would to buy would be like, if I can buy South by Southwest. You know, Ted is for sale right now. Maybe you should buy that. I didn't know. Is that, is that for real? Yeah, it's fair. It's for,

Starting point is 00:11:13 I think it's public knowledge right now. They're shopping it around. And he's going to give it. I think he's going to give it away to whoever is going to be the best new owner of it. So maybe you should go pitch Chris Anderson. Yeah, I mean, I think Ted has done an incredible job of turning an event series into a media company. But they've stopped at media company. So I think this is a perfect, like Ted would be incredible around like, okay, all these people who watch,

Starting point is 00:11:42 TED videos, ideas worth spreading. How do you give them not just ideas, but also tools? And what are those tools and how could you vibe code your... So just so I understand. So let's say you buy a yachting website and it has a big community of people that are all into yachting and they just read the news on there. And you're basically saying you could vibe code a bunch of cool data tools. So for example, what are people paying for their yacht?

Starting point is 00:12:12 and how much are they paying to maintain them? And then once you have that data, you can sell them services and all that kind of stuff. Exactly. Yeah, I think that's really smart. I mean, we've looked at a lot of different media brands and, you know, buying these kind of distressed assets and stuff. And we've always passed because it feels like,

Starting point is 00:12:33 it feels like, do you think about, like, what's happened to, like, Forbes now? It's like, do you remember, like, seeing friends post on Forbes? you'd be like, whoa, like, that's crazy. They're on Forbes. And then you realize, like, Forbes basically just opened it up to anybody and any SEO person could just post on Forbes and just kind of died. Like, I'm really curious. Like, I love Kevin Rose, but like relaunching Dig, I'm like so curious about that.

Starting point is 00:12:57 Like, does that actually work? And is that nostalgia worth something? And I'm fully with you in terms of, like, taking a community that is monetized in the wrong way and just monetizing it in the right way. do you think that's a unique AI related thing though? Because I think that was an opportunity five years ago and it still will be now. It'll just be a little easier to build.

Starting point is 00:13:19 It'll be easier and cheaper to build the tools to sell them. Yeah, I think it was an opportunity back then, two years ago, three years ago, but I think that it's not a little bit easier to build. It's significantly easier to build. And also, maybe even more importantly, there's demand from consumers to try new apps. For example, the flavor right now is Cal AI,

Starting point is 00:13:49 apps like that. Those companies are doing like $20 million AARR on downloading an app. And you know how hard it is to download apps and getting someone to download an app. So I think that what AI has done also has just made it, it's created new use cases. It's created, you know, with CalAI, for example, you take a picture and it tells you using LMs how many calories it is. That's cool. You couldn't do that a few years ago. You know, it's interesting. I got to have dinner with Charlie Munger, like, for the first time, like six years ago. And, you know, he's like, at the time, he was like 97. And we would be talking about all these different ideas for investments and stuff.

Starting point is 00:14:37 he would always just say like, it's hard. Investing is hard. And I was kind of like, you know, it was like 2018 or whatever. Like I was at my peak in terms of like, you know, no one else was doing what we're doing. We're just like on a buying frenzy where, you know, valuations are great. And I was like, what is this old guy talking about? Like, and what I realized in retrospect is that he, you know, he just got such a big scale and such a competitive part of the market where, you know, you're buying businesses for 10. billion dollars to move the needle that it just becomes inherently competitive. And I think that is happening in in all arenas right now. And I think I kind of am becoming, I'm slowly becoming the

Starting point is 00:15:21 crusty old man at the table where, you know, think about like all the things that were kind of easy to start before, like, you know, newsletters and all that kind of stuff. I just think there will be almost unlimited competition in all these arenas. Um, whether. whether it's apps. Like right now, like the example is like Cal AI is really cool. Like I downloaded that app and it's amazing. And there's tons of those. But there's going to be like a hundred of them in the app store in like two months. And then Cal AI will not be able to charge what they are. And I just go like those are good one-off bites. Those are good for going from a college dorm to being a millionaire. But they're not going to get you to a deca millionaire. I think like your

Starting point is 00:16:03 equity value is just toast within a very short period of time. Because I just think, the pricing ultimately goes to zero. And I think as well for like newsletters and attention in general. Like I recently, I own like a local newsletter business and I was goofing around with Lindy. And I was able in about an hour to build a Lindy agent that could go through. It would basically like choose this. I chose a city like a city near me. And I said every day I want you to, there's three different agents.

Starting point is 00:16:36 One sources all the news and it goes through like local Reddit, like forums, local news websites or whatever. And then another one that like takes it all and writes it and punches it up and makes it like fun and cool or whatever. And then the other one that formats it as a newsletter. And it is perfect. It is better than any local news letter that I could write. And it's more detailed because it goes to all these different sources. And it even fact checks itself. So I just think it's going to become crazy.

Starting point is 00:17:05 I think all of our inboxes are going to get overloaded with great content that's written by AI. And I think a lot of these businesses are just going to get so freaking hard. The other thing that's interesting is like when you talk about data, like I find that whole thing really interesting. Because I think data and a network effect are kind of interlinked where, you know, I was talking to someone at one of the frontier models. I won't say who. But I was like, oh, you know, we own all these social. networks or whatever and they have all this interesting proprietary data. And he was like, oh yeah, we probably ate all that.

Starting point is 00:17:42 Like, you know, they just, they've gobbled up the entire internet. And like, yeah, you can go and like sue them. But like even a data, you know, it's not like a data moat because now all your data probably exists within an LLM already. Right. Like it's hard to identify like what is actually unique data because all they need to do is gobble up a big enough sample size to have it be predictable in almost any arena. I mean, okay, let's talk about, I want to talk about a few things because you talked about a lot of things here.

Starting point is 00:18:11 So the Cal AI example, if I was Cal AI, what I would have tried to, so I would have acknowledged the fact that they're going to get a thousand competitors. And I would have said, okay, how do I create a network effect in this app? It doesn't look like they're doing that. I don't know, but it doesn't look like they're doing that. So all the people that have downloaded CalA. are people interested in health and tracking their healthy lifestyle? And maybe they want to lose weight. So does the modern Weight Watchers, which was a multi-billion dollar company that like Oprah was a part of,

Starting point is 00:18:51 does the modern Weight Watchers start looking like a Cal AI in 2025? Probably. But they had to build the network effects to go there. But Weight Watchers started as a physical moat because you'd go. to a meet up in your town and then you'd have all this social buy-in and you have people that are in your network selling you, right? I think Cal AI are these guys just like, I just don't know what my advice would be like, if that guy came to me, I'd probably be like, just sell. Like find someone who doesn't understand AI and just like sell it and be like, look, like the ARR keeps going up.

Starting point is 00:19:26 You should buy this for $5 million and get out because personally, like I tried Cal AI, but I also I think I tried like two or three others. And I think I ultimately found a tool that was just cheaper. And like maybe I'm just a cheap bastard. But like I think they basically get bifurcated into, you know, there's all these like AI photo editors on iPhone. And they're just for suckers, right? Like it's literally like do a seven day free trial. And then it's like $49 every two weeks. And you're suddenly spending like $1,500 a year for like a basic AI photo editor. There's those guys. And they're like borderline criminal. Like I, find that like crazy.

Starting point is 00:20:05 And then there's people that'll just create these things for free or a dollar. And then ultimately Apple will just eat them. And this is the thing is like the largest, at the end of the day, like the people that will ultimately capture a lot of this value are going to be whoever controls the devices. Because I do believe like a lot of these LLM, a lot of this functionality will ultimately be implemented. So I think like if you can find a niche, like let's talk about two different things. So one is buying a business or starting a business that is a five or a 10 year business.

Starting point is 00:20:42 And then separately we can talk about how to make your first million bucks. Because I think like if there's nothing wrong with making your first million bucks and building a vibe coded crappy app and then it disappearing in tears, that's awesome. That's a great win. But for guys like you and me, like it's. got to be something that is sustainable. Yeah. Why don't we talk about, should we dig into a couple ideas and stuff? First of all, before we dig into that, I'm just curious, we should talk about like what

Starting point is 00:21:11 kind of automations and tools we're using. I'd love to hear what you're doing right now. So right now, like currently, I'm spending a lot of time with Gumloop. Have you played with Gumloop at all? So Gumloop, if I understand it's similar to. to like make her lindy yeah exactly it's very similar and by the way like i don't it's not like i'm recommending one platform or another platform like my my mow right now is just to get smart on all the platforms and i'm starting to notice that there's something i like more in lindy and there's something

Starting point is 00:21:50 i like more of gum loop like there's something i hate about gum loop that lindy or someone else does better. So my recommendation, honestly, is just to try them all and see what you like and you don't like. But I'm, you know, my company has been running on Zapier for the last four, you know, four years. And Zapier has been really good to us. And when I found Zapier, it was a game changer, honestly, for our business. We were just automating so much stuff that we had human beings do. So for example, sales leads. We would have someone go in every day, look into our sales CRM,

Starting point is 00:22:36 and basically report back, who are these leads? Is this a good lead? Is this a bad lead? And then create reports. With Zapier, which was really cool, we created a Slack channel, and every lead basically gets scored

Starting point is 00:22:52 and gets put into the Slack channel. And the whole team gets to see the sales lead. So to me, I'm looking at that as like a co-owner of the business. I'm like, this is amazing. But the problem was there really wasn't

Starting point is 00:23:09 much intelligence. Meaning it was more of like the scoring wasn't very good. It's like Boolean, like if this, then that. If it doesn't match perfectly, it's messed up. It's messed up. So what it basically did was it just posted it to a Slack channel, which

Starting point is 00:23:27 With Gumloop, now we can actually go and say, like, okay, this sales lead came in. Andrew at tiny.com. Who is Tiny? What time did he come in? Was this at two in the morning? Maybe people at two in the morning actually want our services way more. Maybe we can charge them more because he came in at two in the morning. What is the size of Tiny?

Starting point is 00:23:54 How many companies do they own? What is the revenue? how many employees and just add intelligence in that process. So from a sales perspective, doing that with Gumloop

Starting point is 00:24:07 has been pretty fun. And do you guys do, so let's say a lead comes in, do you have an AI agent, like email them back and say like, hey, can you tell me more about your company

Starting point is 00:24:15 and stuff? Exactly. That's cool. That's really cool. Yeah, but only if it's not maybe a, let's say we would score you as a 10-on-10 lead.

Starting point is 00:24:25 someone like you actually might not want an agent to reach out to you. It's got to be a fake person and it's got to be delayed a certain amount of time. I know, but I also think that, you know, people are saying that agents are going to automate everything 24-7. There's some things that the human touch is going to be better, right? Like me recording a video, if you, you know, signed up to one of our forums, maybe me actually recording a video and being like, hey, this is Greg. You know, we have a common friend in Sean Puri. I'd love to talk to you about our service.

Starting point is 00:25:05 That probably would be better at this stage with a human in the loop, in my opinion. Totally. But I do believe like pretty soon Greg can be an agent and just be a video that looks like you. Yeah. Yeah, I agree. So I've been messing around with that. That's top of mind. mind right now. And then the other thing that's top of mind, and I want to hear what you've been up to,

Starting point is 00:25:29 is just Manus AI, playing with Manus AI. It's crazy. It's so cool. I just, my, I literally was, I always have this problem where I get excited about a project and then I go on Twitter and I'm like, hey, I need someone to run this thing or I'm hiring for this role. And then I get like 50 emails. And so I did that recently with a project, which I'll talk about later. And I got so many applicants. just dread it because I just, I hate going through all the applicants. And if I hand it off to my assistant, like she kind of, she's busy with other stuff and like she doesn't really know what to look for or whatever. And so I took all those emails. I exported them out of Gmail as a zip and I uploaded 50 candidates to Manus. And then it just spent 30 minutes scoring them all based on what I

Starting point is 00:26:15 needed. And it was like, these are the best three people. And then I just focused on those people. And it was totally bang on. So cool. Yeah, I mean, and it's unbiased too, right? Even if it, I'm not saying you're biased, but everyone has some biases to any of their decisions, subconscious or not. So I think that that's also really cool is, I mean, I guess you can say the LLMs have biases too, though. They have human, I guess they have human bias because they're trained on our data. Exactly. So, yeah, I'll tell you kind of like a sampling of the things I'm doing.

Starting point is 00:26:52 Like one of the dumbest automations I built recently, but I find so satisfying is I have a Lindy agent that looks at my calendar events when I create them and it adds an emoji, which is so simple. But now my calendar looks really beautiful. More impressive than that, I built a Lindy agent that kind of does what you're talking about. So it ingests every single email that I get. And then it labels it or archives it based on context. and I have all these examples of like different kinds of emails and whatever. And then if it's something that I'm likely to say no to, it'll actually draft the no email for me.

Starting point is 00:27:31 And I've tried like superhuman and seraph and a bunch of these other ones, but I ultimately wanted to roll my own and just have it, just know exactly how it worked. And that's been amazing. The other tool I've been using a lot is VSEL V0. If you mess with that. Dude, crazy. Like I own a pressure washing business. And it's like, you know, I run it with some kid locally.

Starting point is 00:27:54 Like it's just for like kind of like a hobby project or whatever. I just don't have time to like dig in and make the website better. But they did this like version, him and a guy that works for me did this version of the website, which is like it's fine, but it could just be so much better. And I just threw into V0. And in like 30 seconds, it had completely punched up the website and we had code for it. Like to me that is just so. mind-boggling. That is something I previously would have spent like a month, honestly,

Starting point is 00:28:26 maybe doing. So that's been really cool. But the thing I really want to talk about and go deep in this episode is MCP. I think it's so fucking cool. I've been implementing a bunch of that stuff custom. And I actually think that might be like a my first million kind of opportunity for somebody to really do MCP right. And I'm curious, are you using MCP for? anything right now? I haven't. No, I'm not. I am not there yet. So, yeah, why don't we dive in? I've got, I've got a bunch of ideas and stuff, but the first one is, so MCP stands for model, what is it, model context protocol. And basically it's a, it's a way of piping in data or APIs into an LLM. So the LLM can basically query your data set. And so I had this, like,

Starting point is 00:29:21 total breakthrough with Claude, maybe like three months ago. One of the things I love about AI is it's like, I find it's like a bicycle for productivity. Like I just get so much more done. And one of the main reasons I get more done is because I don't have to ask questions of other people. So one of the things that would often happen to me would be, you know, it's a Sunday. I'm working. and I want to know, hey, what's happening in this business with our margin? Or how much are we spending on Slack? Or, you know, just these random questions. And those would be things that unless I really wanted to get fiddly and dive into a bunch of accounting or whatever, I would just email to somebody.

Starting point is 00:30:05 I would lose interest in that problem. Let's say like I was cost cutting or something like that. I'd lose interest in the problem because I have to wait four days for the answer. And then I'd find out on Thursday. What I did is I went to Claude and I said, hey, what information would you need from zero to be able to give me like a full financial analysis on my business? And so it gave me a list. And, you know, it took like 20 minutes, but I went and I exported all the, you know, CSV files out of zero. I trained a cloud project.

Starting point is 00:30:35 And it was basically on my accounting for all my personal holding company and my personal life for the last two years. And it was magic. Like just absolute magic. Like at one point I said, how could you optimize my taxes? And it said, oh, you have a credit line over here and you have a credit line in your company. If you use the company money to pay down the personal credit line, you can just immediately write off like, it's like I think it was $100,000 of tax savings or something. So to me, that was like, oh, my God, this is magic.

Starting point is 00:31:08 And when I learned about MCP, I saw that as an opportunity to try and get live data. Because the annoying thing about the cloud project is it's all historical data and it's all got to be manually updated. And so I hired a developer to work with me literally like came to my house and went on my computer and set it all up or whatever. But we got it working and it's really cool. So now I can say it's linked to all my, all of the companies that use certain like QuickBooks or Zero or whatever and I can query. And I can say something like, you know, across all my companies, what's happened. with revenue or how much are we spending on this thing or whatever query it is and it can make graphs and charts and tables and stuff. And it's not perfect. Like I'd say, like sometimes the

Starting point is 00:31:54 numbers are wrong, but it's a breakthrough. And so the opportunity, I think, here is to make the easy kind of one-click way to do this. So an MCP server for the Mac where you can basically say, I want to integrate all these APIs. And then it just manages downloading the data is onto your computer. It manages the interface with Claude. And I know there's a few people that have kind of tried to do this, but I don't think anyone's really done it properly. And a service business around it. Like I might not want, right? Like I might not want to hire developer to do this myself and sort of figure it out. But if someone could go, and probably honestly, charging monthly, you probably can get away with charging monthly for something like this.

Starting point is 00:32:38 Well, the best way to do it would be to be the competitive advantage would be it's all about security. So you're the most secure. You're the default. Like think about like one password is not hard. Like anyone can vibe code one password. What matters about one password is everybody knows one password has not been compromised and they trust. And that's why they pay, you know, I think we pay 200 bucks a month or whatever to use that software. So I think that's a huge opportunity if someone can actually do it properly. And it might be a good one year business. It could also be a good five year business if executed well. I'll give away the, name for this business. So I would brand it as

Starting point is 00:33:14 SMCPs.com. So think of it, think about, because MCP is a protocol just like HTTP. But, you know, you also have HTPS. I think the S stands for secure. So if you could basically own, I'm the secure MCP guy. Everywhere else is not secure and create that positioning dichotomy.

Starting point is 00:33:41 I think that that itself is like going to help you win deals. And I think if you want to own that domain, you got to buy it off, Greg, for $20,000 because I assume he just bought it. No, I do this on the pot a lot where I give away domains. And then within the first like five minutes of posting, someone snipes it. I don't know how you do this, you people, but I don't even care. Like I just, I want someone to go and do this, build it, and pay it back somehow. I like that. I really like that idea. If anyone wants to do that, we should back them.

Starting point is 00:34:17 Not that they need, not that they need our money. Yeah, but, you know, it could be, it could be fun. It could be fun to be involved. To your earlier point, we have distribution. So we actually do bring value to the table. Exactly. We're not just, we're not just cranky old guys. We do bring distribution. Okay, so that's a really good idea. Also, I was thinking about what you were saying about your pressure washer business. and I've got another idea for someone to steal that I think is as simple how to make your first million. So I think that what people should be doing is be using, someone should create essentially an automated agency built on top of V0 that scrapes local businesses and prioritizes the ones that have ugly websites, train it on like modern design practices,

Starting point is 00:35:14 minimalists, you know, make them beautiful. Build the websites and then automatically reaches out to these small business owners and be like, hey, I just redid your website. I love that. Click here to buy. Click here to buy and I'll host it. And then it's just 20 bucks a month to host it. And then any changes you want are $10 every change.

Starting point is 00:35:35 Exactly. I love that. And we have your credit card on file. They're happy because they've got. this, they think they're getting a steal, right? They're like, oh my God, $500 for this website and it's already done and they're going to host it for me? This is incredible. The bar is so low. Like, it's crazy. The services businesses, I mean, there's people making a fortune and they have these terrible websites and they barely do SEO and stuff. And I think there's a big opportunity just to

Starting point is 00:36:01 take best practices and roll them out in that arena. I love that, though. As a first business, I think that's an incredible idea, and it might actually be a good long-term business, at least for a while. Have you looked into web hosting businesses ever? They're phenomenal. They're just like people, it's like an accounting firm.

Starting point is 00:36:22 It's like people just don't want to switch web hosts that often. It's like, look at GoDaddy. Like GoDaddy is the perfect testament to why it's a good business. You can have that horrible user experience in that fugly website and still be like a billion dollar company. Totally. side note i was telling my wife i was like how do you you know how you know someone's in older than 30 is if they want to register a domain and they go to go daddy 100% and they or like you email them and you

Starting point is 00:36:49 go to their website and it's like this this domain is parked by go daddy yeah exactly so okay i got another idea here um this one is like kind of half baked but i think it's kind of interesting so um i was dealing with a difficult person via text message. And I was like, man, like, I really want help with this. Like, I want help. And sometimes I'll, like, take an email thread and I'll paste it into Claude and get feedback or whatever. But I started thinking, I was like, I really wish that I had a way to analyze all my text messages. And somebody posted publicly and said, you know, with these two million context windows, you can now digest five years worth of text threads. And so my idea, so like, basically

Starting point is 00:37:35 like, you know this whole like looks maxing trend? Yeah, yeah. Basically, like, it's these apps where you can like take photo of yourself and it'll like basically hot or not you. It'll be like, oh, you know, your skin is kind of bad and your hairlines receding and like here's all the things you can do. And then you keep taking photos to look better. And so my idea is message maxing.

Starting point is 00:37:57 So on your Mac, and this is just for Mac, I don't know how this would work on PC. But on your Mac, I message syncs to a database. I think it's called chats.DB or something like that. And it's a big huge two-gigabyte SQ Lite database. My idea is you basically have something to ingest that database and analyze it. And it might be with a local LLM because it might be too much of a pain in the ass to upload it. And it basically tells you, it could tell you a bunch of different things, but one example would be it could tell you which of your friends or contacts is exhibiting toxic, narcissistic or psychopathic traits.

Starting point is 00:38:35 Like who should you steer clear of? Who's being manipulative? Another one might be like, how are you exhibiting negative behaviors yourself? Or how are you being kind of like beta or weak in dating and interactions or that sort of thing? And then also like which friends are you neglecting? Who should you catch up with? And I think like some of these things Apple will do over time. Like I think their message summaries is like an indication that eventually they're going to start

Starting point is 00:39:01 digesting more of your messages. But I don't think that Apple would ever tell you who's toxic or narcissistic or whatever. And I just think this is the sort of thing that could go very, very viral. I think this is a one-year business. I think this is easy to copy. I think someone will copy this idea when I share it. They'll make a million dollars. It's not a long-term business, but it's a cool idea. Yeah. So I started thinking about businesses like this where I start with the TikTok. Because you can imagine what the TikTok is going to be, right? Like a viral TikTok, am I a narcissist, showing the screen, oh my God, you know, my boyfriend's a narcissist. Like, you know it's going to work. It's just a matter of figuring out the right format. So I think that I call these

Starting point is 00:39:50 businesses format businesses. So it's it's a game of figuring out the format. I agree with you that, you know, it might only last a short amount of time. But it probably falls into the Cal AI camp. Well, I think it's, I think, so I think, and this kind of feeds into my other idea. So the other, and I think what we said before of like, if it's highly, highly secure, then it can be a high quality business, right? So if you think about other applications, so those are kind of the fun Gen Z ideas, right, of like, ooh, who's a narcissist, whatever. But I think this would be useful for me in terms of business. I think it could be helpful of like who are you,

Starting point is 00:40:33 who are you messaging that you're turning off without realizing it or who's being passive aggressive? Because I do a lot of business over text messages, right? Having someone coach me on how to write more effective text messages is just like how I'm using AI and emails. So like when I write emails, I'm like a, you know, I'll write back an email that just says like sounds good. And some people will view that as passive aggressive.

Starting point is 00:40:57 But if I add a happy face, then it's great. Or if I write, hey, Greg, sounds good. Thanks, Andrew, right? And so I just think something that can tweak what all of your interactions in a positive way, I think could be really cool. But yeah, it's not like a great long-term one. It's kind of, this idea is kind of like grammarly, but more for like social relationships. Well, and just think about what data is in there. Like I think messages are so much more intimate than even email.

Starting point is 00:41:27 And what I'm craving, like, I don't know that Google will do that. And this is another idea I have is like an LLM that actually ingests not only your messages, but your email and then can actually tell you like what's going on in your life. So like when I go to chat GPT, like you can say something. There's a cool prompt that's like going on on TikTok that's like based on what you know about me, tell me what, tell me what psychological insights you can tell me. And so I did it. And it was like, oh, you're you're a searcher. you're always searching for the next thing or, you know, whatever it is. And it's got all these, like, different insights.

Starting point is 00:42:02 And they're kind of like, it's almost like horoscopes, but they're kind of real, right? I actually really, I did this thing recently where I told Lindy to look at my last 250 emails I had sent and then say, what projects am I working on and what's the status of each? And it was pretty crazy how well it was able to summarize everything that I'm working on based on my email. And so I think that, um, I don't. don't know who's going to do it, but I think whoever does that will do very, very well. And I think it's probably Google. But apparently the processing required to do it is quite difficult.

Starting point is 00:42:39 So the idea, the startup idea there would be to build. So what I did is I use Lindy. And so Lindy sends it to an LLM and reads every email. Some people are very concerned about security. They don't want to do that. I think in the same way this secure MCP idea, someone could build a secure local email processor that ingests all your emails tells you what's going on and labels them and does all that on device.

Starting point is 00:43:09 Yeah, I think, and that's why I actually think this could be a bigger business. Like, you can actually create a secure holding company. You know, you only work on secure products and agencies. and service businesses around, you know, AI meets security, probably do pretty well. Okay. Okay, my last idea. So have you ever got a loan from a bank?

Starting point is 00:43:36 Yeah. So, like, it takes fucking forever. Like, whether it's bank or private credit or any of these people, it's like, you know, just to get a call is a week. And then it's another two months of like, hey, can you give us this document or that document? And then all they do is they just like package it up and they send it to like some wonk at head office who like reviews it all. And that is

Starting point is 00:43:59 basically an LLM, right? So my idea is AI based lending. And the idea would be like you choose a vertical. So you'd say like, I want to lend to, you know, SaaS companies or whatever it is. And it's just like, hey, how much do you want? And then it starts asking questions. And it might say, oh, do you Stripe, okay, link us to your Stripe API. Hey, upload your ID. Upload your articles of incorporation. And it basically just automates all the diligence and packages it up. And then at the end, you still have a human who goes and verifies that, you know, there's nothing sketchy or whatever it is.

Starting point is 00:44:37 But to me, this seems like a very obvious idea. And it may already exist, but I just haven't yet seen that. Like, if I could go and get a $5 million loan in 48 hours for like buying a business, that would be incredible. And I don't know why that's not possible other than the fact that it's this very antiquated industry. Okay, I have an idea similar to this I want to share. So did you see at SHL Sahel Lavinia tweeted,

Starting point is 00:45:07 I want to give $100,000 to any vibe coders that will give me 10% profit in exchange for that? Did you see this? No. Went viral. I thought it was really interesting. So here's the idea, a little mix of Sahel's idea and your idea. It's the bank of vibecoding.com.

Starting point is 00:45:29 And it's basically this. It's an AI lending LLM where you take a percentage of, of course, you know, you're not going to accept everyone, right? But there's criteria in the LLM that says like, okay, you know, who are, you know, have you maybe launch successful projects in the past? Have you, do you have more than 10,000 followers on X? Is the project, are the projects that you're going to work on have stripe integration? Of course, you need that. So, and then you just take a percentage of profits based on, right? Yeah, super smart. I mean, any of these, these are the things that like previously someone would say this to me and I would just say like, no, because it's an administrative nightmare and you need to

Starting point is 00:46:19 have an army of accountants and lawyers to do it all. And it just wouldn't make sense. But I think now it totally makes sense to the point of like if you'd pitch me on the website, hey, we'll approach all these small businesses with pre-designed websites. I'd just be like, oh, that's a nightmare business. But with an agent, that's a great business. Totally. Okay. So you started off this conversation, Grumpy, Charlie Munger, and you've now ended the conversation. No, no, no, no. To be clear, I'm still saying almost all these businesses will be zeros in five years, right? Right. But there's a possibility that there's something.

Starting point is 00:46:55 And I think that, you know, I would say there's a 20 or a 10% chance like five years ago, pre-AI of starting a business and having success. And now I'm just like, oh, I think it's like a three to five percent for most businesses. because I think the competition is just so profound. The counter argument is it's so hard to predict even five months from now that how are we going to predict what's going to happen in five years from now? Well, I just think, you know, the Dario Amadei quote of like, you know, all keyboard jobs are gone within two years. Like, if that is true, then, like, I think, like, all businesses are hard, right?

Starting point is 00:47:37 I'm more just going, you know, that maybe that's not a foregone conclusion. I'm just saying hard is coming. Competition is coming. And I think the moment that someone can just prompt an LLM to compete with you. And like, let's say I could go to an LLM like Manus and I could say, hey, Manus, I want you to build a calorie tracker to compete with CalAI and I want you to beat it. And that's all you have to do, which is totally possible in my opinion. I don't see why that couldn't get to the point of being real within two to three.

Starting point is 00:48:07 years. So I just think that's a hard world to be a business person in. It's exciting. It's the most excited I've ever been for business because it's easier than ever to start a business. But the worst businesses are the ones that are easy to start. That's why I think audience brand is really important. Like, for example, bank of vibe coding, that's just sticky. So you and I have personal monopolies, right? One of the reasons I've spent so much time over the last couple of years building an audience is because distribution, like you said, is valuable. And if someone resonates with you or me or Sam Parr or whoever, then they're going to stick around and keep consuming your content.

Starting point is 00:48:48 So I think that's an unfair competitive advantage. Yeah, exactly. Also, with that vibe coding idea, the bank of vibe coding, you're also, if you build it now, you're integrated, right? And if you get a thousand vibe coders, to go build stuff and you own 10% of 1,000 vibe coders businesses, like that might be enough to set you up for life. Yeah, VCB.

Starting point is 00:49:16 It's like SVB except Vib coding bank. I like it. You should do that. I would love to invest in that if you want a partner. I would love to partner with you on anything, man. Awesome. That'd be great. Before we head out, I'm just curious.

Starting point is 00:49:33 I know you are building an app. I think it's called vibe. Is that what it's called? Yeah. So basically, the quick version on it is I used to dress like a total schmo. And then I got divorced. And I was like,

Starting point is 00:49:50 okay, I need to dress better. I hired a personal stylist, which sounds really like fancy, but it's actually not very expensive. And it was great. She'd send me like nice outfits to wear and stuff. But I'd have to literally text

Starting point is 00:50:03 this woman and be like, oh, should I wear this shirt with this or whatever? And so I started just uploading photos to chat GPT. And I'd be like, hey, I want to wear these jeans today. Like, what goes with it? And then as I'd get dressed, I'd just send more and more photos. And it would be like, oh, tussle your hair and like, um, wear this watch that you have. And then I got more advanced. I started prompting it with all the clothes I had. Um, and then I was like, oh, this is like a cool thin wrapper. And so I hired a agency that I know and for, you know, 30, We threw together this app called Vibe. It's in the app store.

Starting point is 00:50:36 Anyone can go look at it. And I'm actually on the hunt. That was the, when we were talking earlier about Manus digging through all the applicants. So I had about 50 people apply. I've got a couple candidates that are at the top, but I need someone to run this thing. So I would love if any of you folks out there listening would want to run this. You have to bootstrap it or raise money for it. I'm not going to put it in any money.

Starting point is 00:51:02 you need to not need a salary, but I'll give you tons of equity. That's the gist. And how do people apply? Just email me. Cool. I won't say my email. They figured out. It's pretty easy.

Starting point is 00:51:14 Cool. All right. I won't include your email on the show notes. But comment below also if what you think of the idea and Andrew, thanks again. Go ahead. Oh, yeah. I mean, the cool, the unique thing about it is you take photos of all your wardrobe and then it'll tell you like hey these are outfits that'll work so you'd be like I'm going

Starting point is 00:51:37 on a date night and it's a summer night what should I wear and then it'll base it all off your wardrobe question on this business idea is this a Cal AI idea or is this a big business in five 100% is a Cal AI this is like you're I want like some college kid who's like I want to get in and get out I want to come in I want to get it to a couple million dollars of ARR I want to cash flow it And then maybe it's a thing in two years, but I have a feeling this is like a moment in time. I appreciate that honesty, by the way. That's a depressing pitch. I just want someone to do something with it.

Starting point is 00:52:12 But I don't know if it's a big long-term opportunity. Yeah. Cool. I appreciate the honesty. Andrew, this is, you have an open invite to come back whenever you want, whenever you want to be optimistic, pessimistic, God, ideas, trends. You know where to find us. Thanks for coming on and sharing some ideas. Yeah, man. Thanks for having me. Talk to later. Later.

The Startup Ideas Podcast - If I wanted to build $1M+ AI startup in 2025, I'd do this

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.