The Koerner Office - Business Ideas and Deep Dives with Chris Koerner - I Told an AI Agent to Make Me Money. It Did. - Ep. #303

Episode Date: May 25, 2026

Check out my newsletter at ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠��https://TKOPOD.com⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and join my community at ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://TKOwners.com⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠━I sat down with Brandon Doyle again and we tested what happens when an AI agent tries to build real businesses from scratch. Brandon showed me how he uses OpenClaw through iMessage, then we talked through a few business ideas an agent could launch, including custom bedtime stories, website outreach for local businesses, and Facebook Marketplace arbitrage. A month later, Brandon came back with real results. His bedtime story business got over $300/month in MRR, but the big win was using an AI agent to build websites for local businesses, send them postcards with QR codes, and turn that into a little over $8,000/month in recurring revenue. You can find Brandon on X at https://x.com/Brandondoyle and learn more about his company at https://getdavid.ai/.Enjoy!---Watch this on YouTube instead here: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠tkopod.co/p-yt⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Ask me a question on or off the show here: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-ask⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Learn more about me: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-cjk⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Learn about my company: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-cof⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Follow me on Twitter here: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-x⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Free weekly business ideas newsletter: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-nl⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Share this podcast: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-all⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Scrape small business data: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠http://tkopod.co/p-os⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠---

We have a little over 8K a month from this experiment. So I wanted to bring you here so we could come up with some business ideas that agents could launch for us entirely. And then I'm going to fly you back here in a month. We're going to touch base on if those ideas made any money and if so how much they made, how you did it, how they can copy you. It worked. And we have like a little over 8K a month in MRR from this experiment. Hold on, hold on. $8,000 a month in monthly.

recurring revenue from this experiment that an AI agent built for you. Yeah. I need to process this. So I actually just have mine on my phone. I just text it back and forth. I named mine Zach Morris. So anyway, Zach just does whatever I tell them to at any given time. It's nice to do it via I message. It found 350 local businesses in Utah. Made the 350 websites. How much did it end up being per website in token costs? A handful of cents. Wow. Yeah. Oh my gosh. Yeah. That one much better than I expected. Yeah. Yeah. Same. This is an effort of using open claw in unique ways, but making real money while showing people what is truly possible. And honestly, I want to have like a handful of things like this. They're making real dollars. Bangor alert, bang or alert,

listen, there's not a coincidence that some of my best episodes are my AI agent made a business for me and make me money. That's cool. Who wouldn't want that? So we re-engineered that today. This video was filmed over the course of a month. My friend Brandon is a baller at AI agents. We kind of brainstorm some ideas of what types of businesses an AI agent could start for us. And then he actually launched some of those businesses. And then he comes back a month later. We talk about how it goes. And you're going to be shocked. Okay, Brandon, thank you for flying to Texas. Glad to be back. Okay. So you've been thinking of some ideas? I have some. Yeah. And are we trying to use OpenClaw? I think so. Yeah. Okay. So for those of you who don't know Open Claw,

OpenClaw is basically like an AI assistant that can do pretty much anything for you if you set it up right, and it can use any AI model that you connect it to. So it could use Claude or Chad or Gemini or Grok, but also like DeepSeek and Quen and Kimmy and other ones from China. So it can use anything. And you can also like give it access to a credit card or whatever and let it spend money to then do other things for you to integrate with whatever. to accomplish whatever tasks you need it to do. And so anyway, I like to push the limits with mine that I have that my company, David

AI, set me up with because, I don't know, OpenCla is amazing and it can do so much stuff. So before we get into some ideas, what have you been using OpenClaw for that people might find interesting? Yeah, good question. So you can interface with it in different ways. So you can interface with it like on your computer, like in a browser. You can do it in chat, in I message, in telegram and whatever. So I actually just have mine just on my phone.

I just text it back and forth. A lot of people like to name their open claw. I named mine Zach Morris. Everybody Bell. I admired him growing up. Me too. So I couldn't think of a name. So I just named him Zach Morris.

So anyway, Zach just does whatever I tell him to at any given time. It's nice to do it via I message. And with the little blue bubbles just kind of makes it feel, I don't know, more personal. It can do anything that AI does, but like more and longer tasks that are more. in depth. So whether it's research, but like, hey, go do research on all the new startups in the whatever home services space and give me the details of like what is their marketing strategy, how much money did they raise if they raise money, what is their moat, what aspect of home services are they tackling, give me all the links, all the details, put it on a Google Doc for me

and then tell me how I can change something that I'm doing, blah, blah, blah, based off of that. So it could do that, but it could also be like more in depth in the sense of, okay, now go and sign up for a demo of each of those companies and tell me what the onboarding process looks like between your demo sign up and when you actually do the call. Wow. So it couldn't do the call for me, but it could tell me like, I want to know the drip campaigns that they get via email. Technically, it could do the call for you. If you connected it to a voice agent. Yeah, I could just have a camera off thing and have. that be on the call and actually I totally could I should do that that'd be weird and creepy but

I could totally do that. Who cares? I guess it wouldn't be weird but it might be a little creepy. Before you continue, I want to touch on the fact that you built it into iMessage. I think that's like, uh, like on the surface, it's like that's cool. That's a cool little gimmick but I think that's critical because it's frictionless. Yeah. And that's an app you're probably in more than any other app, which means you're going to use

this more than you'd use it otherwise. Right. If it were buried in a browser or like a crappy user interface, which means the more you use it, the better you get at it. And the wealthier you get from it, right? Yeah, totally. I don't want you to undersell how critical it is to have it in like where you already live on a daily basis. Yeah, yeah, and that's a good call out.

So we do this for other companies. We do like AI solutions for companies. And when we do like an open claw setup, which there's other versions now you can do Hermes and other things. But open claws is still the best and most powerful in my opinion. Anyway, we try to set it up where they live the most. So sometimes we'll get a client that like their whole day is in Slack. Yeah. In which case we're like, okay, well, we'll just build yours into Slack.

Some is an email. So we'll build it in an email. I'm in my I message, you know, my message is app on my iPhone all day. So I just figured friction-wise, I want it there because, yes, I don't want it to be in WhatsApp or telegram. Like, I don't use those that much. So, yeah. So when you send an I message to Zach Morris, is it getting on your Mac Mini? that you have a display on

and it's like actually going and searching and you can see the cursor move and everything. So it does connect to a Mac Mini, which I have in my office in Provo, Utah, which is online at all times. It's actually not connected to a screen, my Mac Mini. It is just chilling there. Yeah, that's where it works though.

And then it usually delivers me the information. So I connected it to my Google Docs and Gmail and Google Drive and then I message. So it just delivers every. either in my text or like in a Google Doc or Google She or whatever. Okay. Yeah. Hey, please just take half a second and hit subscribe on Apple Podcasts or Spotify or wherever

you're listening to this right now. It would really mean a lot. All right. So what are some ideas you've been thinking of? Okay. So I saw a couple people talking about like bedtime story stuff for kids. So that's one that I was thinking would be fun, partially because I think it would be easy to do.

So basically what I want to do is. spin up a little website that's basically like custom bedtime stories for kids. So I've done this for a long time with my kids. I don't do it as much anymore because they're a little older, but telling them bedtime stories. But it can be cumbersome to do it every night and to do new ones. So what I actually started doing, which did work for quite a while, is I started going through all of the Marvel movies and just I would tell snippets of those of each movie starting with Iron Man One and going all the way to end game and I would just change the character names to be my kids' names and their friends and stuff but that's where I just get the ideas from because I knew the movies well anyway so that lasted

me like a couple years just going through the chronologically through all that stuff and then I would just do like other superhero movies but I mean that's hard and not everyone knows yeah a bunch of movies really well where they can just pull from and you can only do like so many like princess night stories or whatever right so anyway custom bedtime stories for for you, for your kid delivered to your email inbox each night. Okay. At a set time a day. Yeah.

Like every night at 6 p.m. You'll just get the thing in your inbox. It'll have your kid's name. Like you can up. I'll do it to where you can just upload the kid's name and maybe like preferences. Like do they like princesses or cars or fighting or business, whatever they like? So this is interesting because it's kind of like a one-to-one bespoke newsletter. Yeah.

In a way. And so people can upload their kids' names, some themes or whatever, and then they will get specific stories according to their preferences in their inbox every day. Yes. Which is really unique. It's not an app. It's not a newsletter. It's like a one-to-one newsletter.

Yeah. And I want to point out another brilliance here is I think a lot of people would, if they had this same idea, the form factor they would default to is an app. Like, I've got to build an app. I got to build an app. But guess what? Like, people don't want another app. Yeah.

We don't want another app. We have like 50 apps on our phone and we use like five of them on a daily basis. So the fact, the chances that we'll add a six is almost nothing. Yeah. So you're saying let's just embed ourselves in an app where people already are, mail, Gmail, just like you're doing with OpenClaw, right? Which in general, I think that's how people should think. Yes.

Because taking more real estate off of their home screen is becoming more difficult than ever before. So, yeah, you can't compete. So let's just go where they already are. Text or email. And I think people default to app largely because they almost feel insecure about the form factor like, well, it's not special if it's an email. No one's going to pay for an email. We want an app.

Yeah. And they want to feel proud of it. But it's like that's not good business. Yeah. We don't want a problem solved. In this case, the problem is creative stories for our kids. Yeah. Okay.

So I love the idea. I'll make it super cheap. I'll use OpenClaw for all of it. Then I'll make some ads with OpenClaught. to drive traffic to it. And that was one idea. We'll see what happens. So the idea is,

so in vibe coding, you have, like they call them one shot prompt apps, right, or websites, one prompt and it's built. Yeah. Sure,

you've got to go connect it to a domain name or whatever. Is this going to be a one shot prompt business or is it going to take some massaging back and forth? I don't think it'll be that hard. Honestly, I mean,

open claw is really good. Open claw is as good as you set it up and train it to be. So like, Mine is connected to all the models. So I just have to, I'll basically just do a really, really good voice prompt to IMessage. Oh, that makes IMessage even more brilliant. Yeah, yeah.

So I'll just say, hey, make me, whatever. Bedtime story, custom bedtime story apps for kids. It'll be delivered to the parents. It will charge whatever, nine bucks a month. Like, nine bucks is pretty cheap. Use whatever information you need to be able to generate at least 365. unique stories per year, blah, blah, blah, and, you know, spin up the website. I'll just, I'll have to, you know, manually probably buy the domain that we're going to put

this thing on. Theoretically, if it had your credit card or was logged into your GoDaddy or names your name's your name. Yeah, then it could. Yeah. I'll have it create some ads and then. So ads is how you're going to grow it. Yeah.

Because nowadays, building is the easy part. Yeah. So you're saying Opa can do the harder part, which in this case would be paid ads. Yeah. Paid ads. I'll probably just do them on Facebook and Instagram and, you know, meta does the targeting via the creative. So I'll just spin up a bunch of different creatives and go from there.

Okay. See what the MRR can be. Monthly recurring revenue. What is your goal for customer acquisition costs? If it's nine bucks a month, you have no idea how long they'll stay. Right. How much are you willing to spend? Like 20 bucks.

Like, I think, I don't know, that'd probably be fine. but I mean part of this this isn't my main business sure so this is mostly or this is an effort of using open claw in unique ways but making real money while showing people what is truly possible and and honestly I want to have like a few or a handful of things like this to frankly be able to just like show our clients like hey we we eat our own dog food we're like yeah this is this is for fun but look, we made this bedtime stories thing. We did X, Y, and Z. Look how much money. They're making real dollars. I just think I want to have some examples to show like our AI clients. Like, I want them to be able to think outside of the box better as well. Yeah. Yeah. Okay. So I think we should

run with that. Okay. In a month, we're going to revisit and see it if it was built, if you have any revenue and how you did it. Okay. Okay. I'll get it done. What else you got? Yeah. So on OpenClaw, more ideas. Another one I had was, so there are endless businesses that exist, obviously. There are endless like local home service type businesses. Lots of them still either have horrible websites or no website, which is kind of crazy to have no website. So I think that I can use OpenClaw to find enough businesses who don't even have a website. Okay. I mean, that's easy enough so far. Yeah, I think I can do that. I think I would also be able to spin up websites for them.

Sure. With OpenClaw. And then I guess I would just figure out how do I then hit them up to, you know, like, tell them like, hey, I made you a website. You know, do you want to buy it or do you want to, or I'll give it to you for free if you hire us for marketing or something like that. So I can think of how I would end up reaching out to them. But I think there's a lot of potential in this in and of itself. Okay, so let me make sure I understand this. Zach Morris, go find local businesses, right?

That's a key part of this. Yeah. Local businesses that maybe are listed on Google or Google Business profile. Yeah. Google Maps, right? Because we've got to find them somewhere. Yeah. And it's very easy for anything or anyone to see, you know, there's no website to click.

Right, yeah. Okay. So only return local businesses that don't have a website. Do you want to focus on a geography? I'll just do Utah since that's it. where I am. I mean, usually it's easier to sell anything to a local area. Yeah.

If your people are familiar to me. So, all right. And then you're just going to prompt it. Okay, once you have this business that doesn't have a website, use Replit or lovable or whatever to build a simple website for them based on publicly available data. Yeah. And then you're going to use some methodology to sell that website as a service to them. Yeah, yeah.

And to actually make it work, I probably have. to do a lot. So I'll do like I'll probably do a few hundred, which then also means I will need to use OpenClawn a way where the token cost doesn't get out of hand. Because with OpenClaught, you basically use, you still use like your own plans with Claude and Chat and Gemini and the Chinese models and all that kind of stuff. So I'll try to keep token cost minimal because if I'm going to spin up 300 websites or whatever, that could get out of hand cost-wise. So I'll figure that out too. that would cost per website in tokens it depends if you actually deploy them on a even just on a

replet URL so on like a 20 a month plan it partially depends how much traffic they get you could probably have like seven to 10 websites live on replet okay so but that's 20 bucks a month but if i'm going to do it for hundreds yeah it would start to add up yeah and i don't know how many are actually going to want them so anyway i'll figure a way around that via potentially tapping into the Chinese models or doing a website once and cloning it and I'll figure it out. But I think it'll be doable. Okay. I'd be curious if you could split test this and do one set of outreach to businesses that

don't have a website and then another set of outreach to businesses that do have a website. Yeah. Because I've noticed oftentimes businesses that do have a website are more willing to pay for a better website than businesses that don't have a website are willing to pay for any website. Yeah. You know? Yeah. Because the first camp recognizes the value of having a website already.

Yeah. They might be getting leads from it. And you might be able to convince them that they could get more leads from a better designed or better optimized website. Yeah. Yeah, totally. So that could be interesting. All right.

How do you think you'll get their attention? Yeah. So depending on how many I do. I mean, I'll get, for sure, I know OpenCloud will be able to pull the contact info for all these. So I'll be able to get,

you know, like their names, phone numbers, emails, obviously business names too, which I'll need for the website builds. So I can, I could call them and then tell them. I can email them. I can, um, I can do postcards to try to hit them up in more of like a, you know, old school way and try to stand out that way. But, and it could be a combo of all three, which, you know, maybe a combo is the best way. But yeah, because I'll have all their contact info. So I'll be able to get to all of them. It's just a matter of how. Okay. What do you

think you might be able to price this at. Yeah, I mean, if I can keep token costs low enough, which is my goal, then what I would like to do would be you get the website for free if you hire us for marketing. Okay. And what would that cost them? We have more like AI powered marketing services that we offer that start at like a few hundred bucks a month. So pretty cheap. Yeah. Yeah. It's not a big ask. But that could really add up. Yeah. I mean, there's 33 million small businesses Yeah, yeah. Yeah. So we have packages that range from like three to 700 bucks a month on the marketing side.

So I would be fine with saying, hey, you get this website for free as long. You just have to sign it for the cheapest package. Yeah. Yeah. Okay. So kids stories. Uh-huh. And then outreach to business owners with no website.

Uh-huh. Anything else that you might want to try? I saw that Anthropic actually just basically had employees of Anthropic use Claude to, like, buy and sell stuff for them on Craigslist in like whatever their HQ is like San Francisco or something. So I was thinking that it would be cool if I could try to use OpenClaw to do that for me on Facebook marketplace in Utah and see like can it just like whatever buy a pair of shoes off of Facebook marketplace and flip it for like a $3 profit or couches or whatever. Something this may be a little more cheap and like how much could it actually, how successful?

what could it actually be? Just trying to find the arbitrage of, oh, this looks good quality, you know, signal the buy, buy it, and then relist. Or you could even have it, like, compare and contrast the same items on eBay and Facebook Marketplace at the same time. So, like, let's say, like a specific version

of a Jordan shoe from a specific year. Yeah. All right. And it just scans eBay and Facebook. eBay. Oh, this one comes, like, let's say, fair market value is $136. Yeah. And it sees one pop up in Pleasant Grove, Utah for $78.

Yeah. And it knows they're consistently going for $1.28 on eBay. Yeah. Less fees netting out at $103, right? Yeah. So it's like boom, ding-de-de-ding-ding. Yeah. Reaches out within a minute.

Then mose them a deposit. Yeah. So they hold it. Yeah. And then messages you urgent, hey, Brandon, go pick up this shoe. I need someone in the real world to do this. But while it's asking you to do that, it's listing it on eBay, like listing it live. Yeah.

And it might even be sold by the time you go pick up the shoe. Yeah. Wouldn't that be interesting? Yeah. Yeah. Yeah. I know that'd be awesome. Well, yeah, I'll just try that too.

Okay. So three ideas. Yeah. Touch base in a month. Yeah. Sounds good. All right. Okay.

All right. Well, I've purposely refrained from asking you questions over this last month because I want to be surprised. Before I ask you any questions, how excited are you about your results on a scale of one to ten? Ten. Yeah. Yeah. I'm pumped.

Wow. Yeah. Okay. Yeah. Things have been going good. Do you have actual revenue? Yes. What about profits? P word profit. Yeah, I've got it. Okay. Yeah. Guys, tkowners.com. That's my community where people are building businesses. I do AMAs, Q&As every week live. You can ask me anything you want. You can have accountability partners. It's about a thousand people in there building, starting growing businesses. Check out TKO as in the corner office. TKowners.com. So three ideas. I'm going to recap everyone. Yeah. We had a, you know, help business owners build a website with OpenClaw, help parents. tell more customized unique stories every single night through an email sent at a certain time

with OpenClaw. And then we had a find arbitrage opportunities between Facebook Marketplace and eBay or just all buying and selling on Facebook Marketplace with OpenClaw. Correct? Yep. Okay. Let's start with the duds. Okay, yeah.

If any, which of these were done? Bad news first. Bad news. That last one, the buying selling one. I wanted to do, I kind of started doing it. I canned it. But I think it has potential for sure. I just didn't focus on it because the other two were working.

Okay. That's fair. I mean, it's opportunity cost. You only have so much time in a day. Yeah. And I'm not necessarily passionate about buying and selling things on eBay, whereas, you know, I've got four kids, so bedtime stories is cool. And then the other one was making, you know, meaningful improvements to my actual business.

Oh, okay. So did you see any signs of life on the Facebook stuff? Honestly, I didn't give it enough of a go to even like, yeah. That's fine. Because you ignored it for good. reason. Yeah. So let's start with bedtime stories. Tell me like how that entire thing went. We do a seven day free trial thing. You can even get like a free story. And basically you sign up, you create your account, you customize your child's profile. So their name, favorite animal,

favorite color, you know, do they like, do they like princesses or cars or whatever? It doesn't matter. And then we, as we say, you get a magical story in your inbox at 7 p.m. every night. And we charge nine bucks a month if you have one kid a little more for other stuff. Okay, but you're gonna show us the prompts to build all this? Yeah, I'll show you the prompts. I don't know where you want them. We can put them on the screen or put them in the description. Yeah, we'll add them to the end.

Okay. That's really good design. Yeah, this is OpenClaw. Did you give it like something to work off of with design? Honestly, no. It just ran with that. Yeah, just ran with that. And I was like, oh, it looks good.

Yeah. So I like the design. We don't even push the family or legacy yet. We just have the starter. which is the $9 bucks a month. Okay. Which price point-wise, we just chose that because, I don't know, $9. We didn't want to charge like only $5.

So we just chose $9. And then we started doing some ads. So I ended up making one ad, but I just made an ad basically saying how, like, I'm a dad of four kids. It's hard to endlessly think of new bedtime stories. And that's why we've made dreamtales.a.o to make it easy. Yeah. So did you make the ad or did OpenClawe?

So I made one ad. Okay. And then we had open claw, make some like AI, obviously powered ads. Okay. So we had a few of those two running. What type of an ad did you make? Was it still image, video, what? Mine was just a video, just like me talking to the camera.

Like 15 seconds? Short? Yeah, it was probably like 20 seconds. Okay. Yeah. You let Facebook choose the targeting? Uh-huh. Yeah.

Okay. And then, yeah, the AI ads is more just like, you know, pictures of moms or kids or whatever and then the story. No like, hey gen, AI avatars. That's more still image. and stuff. Yeah. We have like a little over $300 in MRR right now. Holy crap. Monthly recurring revenue. Yeah. So like the 33, 34 customers? Yeah. Holy crap. And we've probably spent like $400 in ads. That's nothing. Yeah. If that were a $9 a month software

product, you'd be willing to spend well over a thousand on ads. Yeah. Based on what your lifetime value would be. Yeah. So I think what we're going to do for fun is we're going to refine the story like make the engine behind the story creation a little better. We're going to try to come up with, right now we have it as $24 gets you three kids up to three kid profiles. I think we'll just change it to 19. So nine for one, 19 for like probably up to four kids. Honestly, it's not that much more work.

Well, up to six kids. They need kids that are young too. Yeah, yeah. So the chance of people even have more than two kids that are underage is pretty small. Yeah. And then kind of decide if we want to do the, physical book option. Maybe just have the wait list for that and not decide on that. But yeah, then dial in some ads more or just better, spin up some more creative and then see,

yeah, can we take this thing? Because basically, you know, essentially we're basically break even off of it so far, but it's MRR. So it's going to continue. And yeah, we're going to lose some and hopefully gain more. But I want to see like if we spend another thousand dollars on ads, couldn't we get another thousand an MRR? Well, that's pretty sweet. Yeah. It pays back in the first month. Yeah.

Which ad perform best? Yours? It was actually mine. But I think it's just because like, okay, they saw a human. They saw a dad. Right. They can. But you could use HeyGen or whatever to do that.

Yeah. It wouldn't be as good, but they're pretty dang close. Yeah. Yeah. It's essentially a SaaS product. It is a software as a service product, but it's the delivery vehicle is just an email a day. Yeah.

If you were to. scale this at any point would you want to kind of take it over and hand it to a human would you want to prompt it better what does scale look like for this yeah no I think probably the best version of this would be it's founder led by a mom who happens to be a creative writer and she is potentially using AI behind the scenes but she's also like constructing the stories or at least like the rubric for a lot of the stories herself so people are connecting more with like oh yeah She's an author. It's a mom.

I just think, I mean, most of our buyers are going to be moms. So I think that that would be how we scale it. Yeah. Yeah. Man, that's exciting. That's a huge one. You literally used an AI agent to build a business for you, soup to nuts, from idea to revenue. Yeah.

So yeah. So that's going good. And, you know, maybe next time I see you, I'll have more to talk about with how, you know, we've scaled this. And then I can talk about the website thing. Mm-hmm. Okay, so here's what we ended up doing. So we used Zach Morris, my open claw.

It found 350 local businesses in Utah, some of which who had websites but were bad, some of which who didn't have websites. So a little bit of both. Uh-huh. And did you ask it to do that? Yeah, I asked it. Yeah, just asked it to do that. You were just kind of curious, kind of like the split test we talked about.

Yeah. Okay. And so it found the 350. It found the contact info for all of them. So like, name, company name, personal name, email phone number. Now let me get specific. Did it just pull the phone number from the Google profile? Yeah, it just did that.

Okay, it didn't go to like Manta or like state directory websites. Yeah. So basically it scraped Google business across Utah for various different services. It ended up being more like landscaping and HVAC is kind of. Well, that's just like it probably just took like a random sample size. And most home service businesses are HVAC, landscaping, plumbing, electrical roofing, whatever. And so it did that. Like that was step one, right?

It did it. It did it successfully. Step two was, can you make websites for all these? I asked it that. It said, yes. It asked me what platform did I want to use. I said, I don't care. Just keep my token costs low.

Okay. It said, okay, where do you want to host these? I said, I don't want to actually host them live yet. Just put them on test URLs. So they are public, but not in like a custom domain. Yeah. So it made the 350 websites. And it made them like with like the Claude API or did it go to Replit?

Yeah. No. So it ended up using more like Chinese models. Okay. Because I said keep token costs low. How much did it end up being per website in token costs? Oh like a handful of cents. Wow.

Yeah. Yeah. Holy crap. So you're doing a few hundred. Here's some of them. Here's some of the sites that it made. So landscape, electrical site. What was a plumbing one?

A roofing one. So yeah, just home services across the board. They're all most trusted. Yeah, yeah, that's true. They're all very, very trusted. All of my daughter's friends are her best friends. Yeah. Okay, so it spun up all the sites and you can scroll and get some more info,

which really just had to pull from for a lot of these. It didn't have it. Yeah, pulled from Google. But, I mean, it's solid, right? It is solid. Yeah. You mentioned postcards, I think. Anyway, we did the postcards thing.

So I said, can you basically send all the, these postcards with a QR code where the owner would then scan the card and then see their site. I want the QR code to point to the site. And then on our on the QR code, I wanted to have the information of my David AI company where they could then, you know, call us to like buy the site basically. I asked if it could do that. I didn't know. And it said, yeah, you can do it with lob. This episode is not sponsored by Lobb. I've never heard of Lobb before. It's is it like a direct mail API? Yeah, yeah, it's lob.com. Okay. So anyway, three letter domain. They must be legit. Yeah, yeah, true. I do know direct mail is like a well-established industry

and there's a lot of technology involved. And there are APIs that you can use for this stuff. Yeah. Okay, so here's the postcard it made. This is the general design for all the postcards. It says, we built you a free website, scan the QR code to see. your new site. Your business info, reviews, and a click to call button all live and ready to go. So that's the front of the postcard. And on the back of the postcard, it says, your free website is live. We built it for you. And then our contact info. Okay. So you scan it. You then go, it would then take them, you know, to their site. And then Lobb gave us analytics, which showed of the 350,

341 of the postcards were successfully delivered. Okay. So it gave us like analytics of that. So I want to say it was like 20% of them scanned. Okay. Well, that's not bad. Yeah. Okay.

So 20% scanned. So I have one like part time sales guy on my team. So I gave him the contact info of the 350. Okay. And I said retargeting. Yeah. So I said, start calling on them. But as soon as we hear that they scanned, for sure, call on those like super hard. Yeah. And then tell them that their site is free if they hire us for marketing and then just pitch our cheapest package. Yeah. Because we do, yeah, it's like a few hundred bucks. Anyway, it worked. And we have like a little over 8K a month in MRR from this experiment. Hold on. Hold on. Hold on. Eight thousand dollars a month in monthly recurring revenue from. this experiment that an AI agent built for you. Yeah, that essentially they did every single thing of except for they didn't close the final deal, which was my part-time sales guy or me.

I need to process this. Listen, I need more people like this to interview on my podcast. So if you know of someone with a side hustle or a business that's unique and cool and super profitable, email Molly, M-O-L-L-L-Y at co-founders.com. That's one word co-founders.com. Molly at co-founders.com. Molly at co-founders.com. Tell her your story and we'll give you 100 bucks if we end up interviewing them. So 68 people scanned. 340 delivered successfully. 9 got lost in the mail. 68 people scanned. I want to know from your sales guy, how many of your closes came from the scans versus the non-scanners? Yeah, it was like 90% scanners, 10% non-scanners. It was mostly from people who did end up scanning. So are they all paying 300 bucks a month or are some paying different packages?

Some are maybe the biggest ones like $1,200 a month. Okay, how many unique customers does that $8,000 represent? Yeah, probably like 20. Okay, so 400 a month on average. A couple do the expensive one. Yeah. Most everyone else does. The cheaper one.

Yeah. 20. So probably more like 17, but yeah. Okay. Yeah. Okay. Two or three of them closed because of the call. they may not have even seen the postcard

just because it was delivered doesn't mean they saw it they might have closed without a postcard you probably don't even know that yeah we wouldn't yeah but the of the 68 people that scanned it they showed enough intent to say yes to the call

yeah so basically if we were to do the math here of the 68 like 14ish of those closed 14 of the 68 scanners 20% scan 20% of the 20% by yes because of the cold call The warm call. Yes.

But I do think, and this I don't think should be neglected, is I think that the uniqueness of them actually seeing the QR code postcard to their website, I think that already signal enough like, oh, these guys are like legit. Yeah. They're thinking outside the box. I like that. Because most business owners get like endless emails a day of like, oh, I can get you number one on Google or I can, you know, run your Google ads or whatever, right?

that's why I was like, well, let's just stand out. Like, yeah, could we use Outscraper and other things and like, you know, just put them on cold at email outreach? Yeah, we could do that. But now we're just a yet another email. Low friction. Okay. I got a little soapbox here. I got to look directly.

I got to look directly at the camera. So like, I think one of the magic things here is you merging the online with the offline. Okay. The more AI grows, the more we need that pendulum to swing towards offline stuff. Yeah. Because the easy thing to do is to mass text people, which is not compliant. You will get sued. You will get fined, right?

By the FCC. Or to do an AI voice agent or to call the emails or whatever. You did the harder, more expensive thing. We'll get into what it costs in a minute. But that makes all the difference because it signals to your potential customer that you guys are real. All the people in their inbox are not real. There's low stakes. There's low money.

There's low investment. Mr. Beast's first viral video ever. The concept for it was, I'm going to give $5,000 to a homeless guy. Right? And see what he does with it. He was going to get that $5,000 from their sponsor. But Mr. B said, make it $10,000 and the video won't do twice as many views. It'll do 100 times as many views because that additional investment conveys to the audience.

They see the work. They see the money. And it just gets them to click, watch, share, and get to the end. They're much more likely to do so. Yeah. And so he twisted their arm. They gave him $10,000. and it got like over 100 million views,

which was his first real big viral hit. Yeah. So that same first principle of human nature is conveyed here. Yeah. Your, like that postcard cost how much each? It was like a little more than a dollar each. Okay, so you spent some hours of time.

This wasn't like a five-minute thing. Yeah. Not counting the calls, pretty little. Okay. Yeah. You spend some time. Yeah. And you spent hundreds of dollars.

Yes. Right. Including the token cost. It was still in the hundred. The token cost ended up being like 10, 20 bucks. It was pretty low. Crazy. Yeah.

And then, yeah, $400-ish dollars on the actual postcards. Yep, which is about, you know, $380 more than cold emails would cost. Yeah. So your potential customers saw that and you stood out that way. Yeah. So I just like anything in life, you got to have, there has to be stakes involved. There has to be risk involved. There is a risk to not taking a risk, right?

Like if I'm afraid I'll get hit by a car so I stayed in bed all day, then I'm going to risk like getting fat because I'm not moving around. Yeah. So it's not a matter of avoiding or even minimizing risk. It's a matter of picking your risk. Yeah. Because there are no solutions, only tradeoffs. Yeah. So your risk, in this case, was $400.

But had you done the same pitch with the same copy to the same 350 people over email, you maybe close one. Maybe. Yeah. But instead, you close 17 to 20 of them. Well, and off of that, like, what the easy version of this is I do scrapers. I set up cold email campaigns and I see how many, how many people reply and I try to close them over email.

Like, that's the easy thing that people. want to do because they don't want to pick up the phone and actually call. Yeah. Because that's hard. Short-term easy equals long-term hard. Yeah. They're picking the easy route, but they're making their life harder as a result. So it's like, okay, how can we stand out? We doubly stood out. The postcard thing and the actually calling. Like we want to talk to you. And actually making a website. Yeah. We made the site too. Yeah. So it's like we're trying to cut through the noise in a variety of ways, which then also, which probably helped our close rates be better than normal, that shows them, I'm, well, these guys, they made me the site.

I got the postcard. They called me on the phone multiple times until they got a hold of me. Like, I want someone like, this is me assuming what they're thinking in their head. I want something like that in my corner. Yeah. Who do I not want to my corner?

The guy that's just emailing me endlessly or messaging me on LinkedIn endlessly saying he can get me more leads. Yeah. Which obviously looks like an AI bot that is sending me those messages. It probably is. Yeah.

So like if you compound your difficult thing, compound your friction full things, then you compound your wins. Yeah. Three hard things. Yeah. Cold call them, awkward, intimidating. Everyone hates it. Yeah. Physical postcards, cost real money, takes time. Yeah.

And then actually building the website for 350 business owners. Yeah. If you only do two of those things, you might get four closes. If you only do one of them, you might get one close. Yeah. If you do three, it multiplies. It's expensively. You do three, you do 17. If you find a fourth hard thing to do, you might get 72.

You know what I'm saying? Yeah, yeah. Like if you show up at their door, the day it arrives, which would be weird and awkward, but like possible. Yeah. I guess the fourth could be you hand deliver the postcard. Yeah. Yeah.

That could get you 70 closes. Yeah. Which still... Then you can actually just chat with him right there and force the chat. Yeah. Now, do you have the data on how many of your closes came from people that had a website versus didn't? Unfortunately, no.

I messed up there. I should have figured out a way to track that better. But once the QR codes were sent out, we basically, lost that part of the funnel. So, oh, yeah. Yeah. So we're going to start doing this basically once a month, now that we saw that it worked, we want to do a version of this experiment once a month and refine it.

And so, yeah, I think next month what we're going to do is only people who do have websites, but who maybe don't rank well in Google will be the thing that we try to do. Like a CEO agency. And still, but ideally still they have bad sites. So I'll spin up a new site. but then point to say like, hey, we'll give you this new one, which will be better for SEO. All you got to do is call us back or whatever. And then we'll try to do the same thing.

Oh, my gosh. Yeah. And then we'll cold call them. And cold calling is not fun. If anybody's ever done it, my first job when I was in college was cold calling. I absolutely hated it. Yeah. But, I mean, if you put in the work, you can make, obviously you can make money from it.

I don't understand where you're sitting here. Like, if I were you, I would be mailing 33 million. Yeah, yeah. We just got to, we do care about quality of work that we put up for our clients. So we need to make sure that we can scale appropriately. Oh my gosh. Yeah. That one much better than I expected.

Yeah. Yeah. Same. I thought it was going to work at least a little. I didn't think it was going to work this well. Yeah. Yeah. Where can people find you or your businesses?

Yeah. Find me on, on X, Brandon Doyle. And then my website is get david. And we do, we do all different kinds of stuff with AI. So we'd love to chat. For business owners? Yeah. Okay.

Yeah. Thank you, Brandon. Okay, thanks. Hey, guys, if you're still listening to this, it's probably because you haven't had a chance to take your AirPods out, you're still mulling the lawn, you're still driving, what have you. If you're still here with me, I would really, really love and appreciate a five-star review on Spotify, Apple, or wherever you get your podcast. It would mean a lot. If you want to go the extra mile, share this episode with a friend that might have an interest in starting a business. It would mean a ton. Hope you have the best day of your life today.