The AI Business That ChatGPT Might Destroy

0:00Welcome to Acquisitions Anonymous. This is Michael. Unfortunately, my camera is broken. So if you're on YouTube, I'm really sorry. Hopefully you see a picture of me talking. Today's deal was me, Heather, and Mills talking about something that we haven't done before, which is in the AI space. And believe it or not, we came to a very interesting conclusion about a business that was only offered for a bit over one and a half times earnings. So it was quite a deal, quite the deal. But we thought some interesting stuff about it. Stick around for the whole episode to see where we ended up. Here you go.

0:30We'll start Acquisitions Anonymous. Hello, another episode of Acquisitions Anonymous. We don't have 100 % beers anymore. Thumbs down on just the plus inventory. When you close a deal, you inherit the cyber and IT problems that have been piling up for years. Enzo Technologies, that's I-N-Z-O, helps acquisition entrepreneurs evaluate technology risk during due diligence and manage cybersecurity, IT, and voice after close. CEO Nick Akers acquired Enzo and recently led the company through its first add-on acquisition. He brings a buyer-operator's perspective to what needs to be uncovered before the wire and stabilized once ownership changes hands.

1:13Enzo offers a complimentary IT risk audit of your target company to help surface the biggest risks in plain English and deliver a 30-day plan to reduce exposure and lower the risk of downtime. visit enzotechnologies.com slash ETA and mention acquisitions anonymous that's I-N-Z-O technologies.com slash ETA I have good news for you guys tell us please anybody want to know what it is yeah yeah my camera's not working so you don't have to look at me oh I thought I thought you were saying this before we hit record Mike Oh, no, I haven't recorded. That was the good news. That was the good news. All right.

1:54Congratulations. I was hoping for better, but. Oh, thank you, Heather. What's new with you guys? Heather, you have such nice flowers in the background. Was it your birthday or something? It was my birthday. That was my birthday, flowers. So I got older. I guess that's good. Happy birthday. Thank you. And I have a vacation coming next week. So I'm looking forward to that. Where are you going? really it's a staycation we do it every year a beach house in laguna so super nice super fun uh and mills are you in columbia south carolina or green i am i know you sent me that article on greenville i love greenville greenville gets all the love greenville and charleston and we're just right in the middle why does columbia get no love because we're the state capital i'm very elitist way of looking at it is that really why well so what comes with the state capital is you're full of federal government buildings and state government buildings.

2:53So there's this crazy stat inside the city limits of Columbia, like 77 % of the parcels don't pay property tax because they're either churches or state or federal and they're exempt from property tax. And so it kind of drags a lot of the like economic development down and everybody else has to pay the property taxes to make up for that. But Greenville and Charleston been a lot more progressive in terms of revitalization of their downtowns. We're kind of catching up now. Did you guys see there were some articles going around where they've gone to these small towns where data centers have been built?

3:30And if they're high property tax states, basically like these towns are sitting on just like tens of millions of dollars a year in property tax revenue and no place to spend it. It's like a fascinating deal. The same thing actually happens out in deep west Texas. are you guys familiar with this in oil country? No. So part of the reason why Texas can Texas is because they make so much money from the oil. So that's why we don't spend much money on schools. Social safety net is basically like, we run these huge budget services because they're tied into all this oil revenue, which property taxes happen that way.

4:10And so if you go out to West Texas, A, it's not a great place to live. So there's not many people there, But the people that do live there, there ends up being these massive income streams for the local municipalities and counties because there's all this oil value and then nobody there to spend it on. So you'll go out and there's these little tiny high schools, like 200 people, and they have$30 million football stadiums. It's crazy. It's crazy. So speaking of AI, I brought an AI deal today. that was i've done some good segues before that segway there that was not intentional that was just dumb luck uh cool can i pitch you guys on this one please it's hot off the presses because i got it via email this morning and i was like we should talk about this one uh it is an artificial intelligence digital assistant that is an on-demand ai agent with low churn cross-channel memory integration subscription-based and does natural phone conversations and they're asking a half a million dollars for it.

5:15The good news is I think they put every AI buzzword in there. I was going to say, I don't understand, but it sounds really good. It sounds like they hit all the high points. It's like every AI buzzword except for organic and all natural ingredients. So, all right. So sales asking price is$500 ,000 for this AI business. Sales are$1.4 million and they made$340 ,000 on those sales. and they're asking 500 ,000. So that is, what is that multiple? 1.15, 1.2? No, 1.5, basically. 1.5 times earnings. Website closers presents an artificial intelligence subscription-based business that has carved out a unique niche in the fast-growing digital agent market.

6:03This agent is not another basic AI writing tool. The platform can schedule meetings, review calendars, send emails, conduct research, prepare presentations, create images, draft long-form documents, manage reminders, and support strategic conversations. The product is designed for business owners, coaches, consultants, advisors, and other professionals who want an assistant they can use from a computer or while they are away from their desk. The seller owns the core code base, prompts, workflows, memory architecture, product design, operating methods, and related intellectual property. No developers, contractors, investors, or outside parties hold rights that would prevent these assets from being transferred.

6:37The acquisition includes the application environment, user dashboard, knowledge base, source code repository, cloud accounts, communication systems, workflow logic, prompt libraries, fund-based support channels, website assets, and related social media accounts. It operates through recurring revenue subscriptions and token-based usage. Their current pricing includes one active plan at$95, while average customer value is approximately$90 per month. The company has also sold higher-level access for much as$500 per month when the product is offered as a more advanced AI agent. The business is not seasonal and has no major customer platform or vendor concentration, and their multi-model setup allows the product to work with several leading AI providers instead of relying upon one model or service.

7:19All right, I'll pause here. Who wants to take a summary of what this is in non-broker language? I'll try because I've been playing around with AI agents and projects. It's just somebody has built an AI agent that is useful to small businesses. And I think they're trying to sell it to folks who don't know how to build them themselves. selves. But when I, you know, it seems like it's got a lot of features to it. You know, they've got the knowledge base. They've got, you know, some phone communications. It sounds like it's sort of a general tool for a small business owner, and they could use it in a multitude of ways with their business.

8:05But I have to say, probably could build one yourself for less than$500 ,000. So I I guess the interesting part of this is this is not a very old business because this capability hasn't been around that long. And we're in a very rapidly changing environment with AI that makes, you know, self-service easier and easier every day. So, but I think it's just a, you know, someone built a tool and made it very generally applicable for small business owners and they're selling subscriptions to the tool. I'm like a total newbie when it comes to this and I haven't really done as much experimentation as you guys have but isn't this basically I mean it seems like they found a way to do something that you can do on your own with AI but they've just now packaged it to somebody and said hey look we've put all the prompts in for you ahead of time.

9:01But part of like the benefit and part of the allure of AI is that it you don't have to code it yourself. You don't have to connect all the dots. You just tell it what you want. And if you're doing the prompts the right way, it is connecting the dots for you. Right. But I think the issue probably with small business owners and really any other sector is there is only a small percentage of people who have learned how to use the tools themselves or are not intimidated to at least dive in there and try. and so this is sort of bridging the gap between the you know the people who have become at least somewhat adept at using them themselves and those that have not and maybe this is sort of a middle ground where you can get people using the tools without having to build them themselves and get a little hand holding probably on you know what it could do so I think that's an that's an interesting approach and I think that's something that probably can last for a while because it is going to take a long time for people to fully come up to speed on all these new tools, especially when some of the language is really intimidating, you know, that's used.

10:10So I think that's, it's interesting, but I think what's really sort of perplexing to me is why buy this business? If you know enough to want to be in this business of selling an AI agent subscription, you know enough to build one yourself. Not, you know, I can see why the end user needs to subscribe to this, but I can't see why someone would buy it for 500 ,000. The next logical thing to me is like, okay, even if like it's a buy versus build question, even more so than that, to me, the question becomes, it's just so much easier for somebody to come compete this away from you. You know, like the threat of substitutes or the threat of alternatives is incredibly high, whether you buy or build.

11:01Like, this seems like it's not differentiated at all. Marketing would be the only way to differentiate this. To your point, you could cut and paste this deal teaser over into Claude and have the bare bones of building the competitor product. But I think the only moat that you could have here, potentially, and I didn't really see it or hear it in this teaser, would be great marketing. You know, if you had some fantastic marketing that would be hard to replicate, that would be sort of where the enterprise value of something like this would lie in my mind. So I would, to push back on that, I think there's actually another boat here that makes this somewhat stickier than it would seem at first glance.

11:46So to make the argument for that, Mills, do you understand like how this works kind of technically, like in the world of open AI and Anthropic? or would it be helpful to kind of talk about what this guy or this person did? Let's talk about it. Okay. So there are, and Heather, correct me if I get any of this wrong, please. So there are fundamental, like the core kind of primary models that are out there. And those are the ones that are like, you hear Claude or you hear ChatGPT, Copilot, right? That is where some big hyperscaler like that has bought at least a big data center. they're running these huge models with 15 ,000, 20 ,000 GPUs, and they're throwing workloads at it over and over again all the time.

12:31And that's where you hear about Claude, Fable, different models of Claude and different models of ChatGPT. Meta has their version, and then there's a bunch of open source ones. But those are kind of these frontier models that are out there. And so those live in these big data centers. They only really work super well if they're on this big-time hardware. So that's why most of us, including myself, or just rent time on those and it's all done in the cloud. So the way something like this works is this is typically what this guy built is called a wrapper. It's like a wrapper application, which basically just gives a basic level of instructions.

13:04But in the background, the models that are running all this are ChatGPT or Claude or whatever. And there's a tell that this is happening because if you look at the profit margin, it's like, wait, this is a SaaS business and you're running it with one guy who only works 20 hours a week with no staff and you're only making$340 ,000. The reality is that other million dollars is going straight to OpenAI or Claude or Microsoft or whoever. So what's interesting is— And so your cost, really quick, Michael, your cost of goods in this case is like token utilization. And you're just— 100%. It's throughput.

13:39You're a toll, in essence. 100%. You're marking up the existing chat GPT or R-Claude with your own wrapper around it to make it consumable. So I'll pause there. Heather, did I get anything wrong or say anything confusing? No, I think you're exactly right. They just made it simple. They used all the tools that are available and are expensive to use and made it really simple, like you said, a wrapper, so that someone can easily come into this and do all these cool functions that they've built. So I think what's interesting and what is a cool feature about this, and you'll see it if you use the desktop app for Claude or Codex from ChatGPT, they actually start to build up a memory of who you are and what you do over time.

14:27So they'll answer things tailored to knowing more about you as a person. Like I asked at one point, I asked for an itinerary to drive across the US. And three days later, when I brought up some question that wasn't really related to that about the types of flowers found in rural New Mexico, it told me in the context of knowing I was doing a road trip. So that's the kind of power. So to tie this all the way back to why potentially this is stickier than we kind of maybe imagine at first glance. These agents that you're running on your desktop when you buy the system are probably learning a lot about what you do, who you work with, and the context of your entire environment.

15:06Such that if you switch to somebody else or a different version of this once you're already a customer, potentially the new version has to relearn a whole bunch of new stuff to get up to speed on your life. So that's why I wanted to talk about how these work, because the fundamental model is having this memory means that something like this can actually get smarter about you as you use it, which helps with stickiness. You're right. And I think that's a very good point that I didn't think of. What's tricky about this is it's got to be so new that you have no customer churn data. You know, you would just have sort of a customer churn hypothesis, like what you just explained, but not really historical data to rely on.

15:46It's also interesting to think about where this is in the life cycle of AI adoption. Like the people that are using AI now are like me and Heather, right? totally comfortable being at the forefront of this stuff. It's okay that it breaks all the time. That's it. The vast majority of the people are where you are, Mills, which is just like, hey, I'm up on a roof. I ain't got time for this, which I appreciate. Hey, everyone. It's Bill. And I want to talk to you about Quiet Light Brokerage. I was so psyched when Quiet Light agreed to sponsor the podcast because I am a customer. I have used Quiet Light to sell three businesses.

16:19And if I were selling an e-commerce or a SaaS business, I really would not consider anyone else. like I said I went back to them three times I work with three different brokers at Quietlight had a great experience all three times even on one occasion they found a buyer for a business that I just didn't know it was even going to be possible to sell so they have pulled rabbits out of their hats several times for me just they've been in the e-commerce and SaaS business brokering game a very very long time they really know what they're doing they have great reach with both buyers and sellers And the other thing I really love about Quietlight is all the brokers there are former operators.

16:54So you can't just show up and go, hey, I'm a lifetime business broker. I want to work at Quietlight. You have to be a former operator. So they all know what it's like to be in the operator chair. So if you go to Quietlight.com, they have free business valuation calls, which they'll do with you. No obligation. Just tell you what they think about your business, what they think it would be worth, and then what you might need to do to kind of get it ready for market. those guys over there are great great sops great systems i just felt like i was really in really good hands all three times with quiet light so if you're interested in selling your business especially in e-commerce or sass hop on over to quietlight.com fill out their onboarding form for a free valuation call and you can tell them bill or acquisitions anonymous sent you it's a question of okay if you want to keep growing this thing like how are you going to make it accessible to the mid-market and the mainstream because like you're not out here mills looking for something like this heather and i are like cool we'll do it ourselves like where does it go from here that's a big question mark for me yeah right i like i said this is sort of for the middle ground people who are ready to use it they can they hear about it but they don't want to build it themselves they don't know how they don't have time for that for 95 a month they get access to something that's already been built that's that's great but it's great for now that's what i worry about it's great for now, but it really will get easier, even though, yes, like when we play with it right now, it does break a lot.

18:14It gets a little frustrating. It's going to get better. You know, it's going to get even easier for people to just sort of, you know, walk through a few steps and have it know about your company or whatever it is you're trying to do, your project. And so that's what, it's really hard to predict anything in this space, what the lifespan of the product will really be, or how many times you'll have to evolve it, I guess, is if it's going to survive, how many times would you have to evolve this business, this tool, and even maybe your marketing approach to still have a business in five years? it seems like from the outside looking in there's like a a high amount of variability or like fickleness between different models and like you know right claude came out with was it optus which was the one that they they like released and they're like actually we can't use it for a few weeks and then they re-released it yeah oh the next one yeah that's right there's another one they still haven't given us yet.

19:16Yeah. But like, it seems like what I hear and see on Twitter is people being like, oh my gosh, you know, Claude just came out with this thing or ChatGPT just came out with this new model within theirs. And now it's like 10 or a hundred times better than the previously best one. And so it seems like they at least acknowledge that they kind of have a certain level of agnosticism when it comes to it. But I wonder how that actually works in practice, given the memory kind of dynamics that you talked about, Michael. Yeah, and you can create that memory so it kind of lives at the wrapper level, which is definitely doable.

19:59There's lots of innovation kind of happening around that. You know, these guys advertise it a feature that they can switch between models, whether it's chat GPT or Claude depending on who's giving the better deal of the week or some of the other ones as well. But again, it's tough to know at this point because things are so early. Is that a feature or a bug? Like you don't know. I would really like if there was a niche here if you sign the NDA and they were like, they acknowledge they don't have customer concentration or vendor concentration. But I think they said this works for business owners or advisors or coaches or something like that.

20:41I'd just be really curious, like if you sign the NDA and like, they're really dialed in on like, you know, orthopedic, you know, surgeons or, you know, health practitioners or dentists or financial advisors or like if they were like, yeah, you know, the guy who started it had a core business and he just kind of grew this alongside it. And that was his niche. I think it would be really interesting because I think it would make for a stickier experience than just, hey, we can do this for landscapers and interior decorators and brick mason. Like we can do it for everybody generically the same. Yeah.

21:19So the alternative, if you're, like you said, this is sort of a generic tool that they can maybe sort of customize once they start using it. But the alternative, if you're a small business owner, is you could hire a freelance person to come along and build what you need, let you go off and use it, come back and customize it some more. So I think that's, to me, that's sort of the competitor model to this. Obviously, there's limited supply of freelance folks out there that you can hire and that are reputable and all those good things. But, you know, I think that's probably one of two solutions that small businesses would look at today rather than just try to do it themselves.

21:57They're going to have freelancer help them or plug into a tool like this. So it seems interesting to me that this is, in terms of like the actual tasks that are being identified, I think it says it, maybe Michael, oh, here we go. So schedule meetings, review calendars, send emails, conduct research, prepare presentations, create images, draft long-form documents, manage reminders, and support strategic conversations. To me, it's like a question of how are people using this and i think like where i am and where like most of my contemporaries are is we bought like a new semi the other day and my truck driver is like taking facebook marketplace listings and saying hey what about this freightliner cascadia you know this year this make and model and chat gpt is like very quickly and this is this guy is like a you know borderline boomer but he loves it and like chat gpt is becoming his friend because it's like well here you know usually around this many miles you have transmission issues or you have fuel pump issues or whatever and i think most people are using it for just one-off tasks i i need help making a decision about this truck purchase yes exactly yeah and so it almost seems like this is just an assimilation of those tasks, not like a complex hierarchy of, you know, related tasks.

23:24It's just more like kind of send an email, respond to this client communication, like whatever is the gist that I get from it. Look, here's a, and I think there's a big problem with this business that we haven't really talked about. And it's just what you mentioned, right? You see all these people using a generic thing. Like this comes across as totally generic. I think that's where it is. And by the way, I did the math. They have about 900 customers. Like, not that many. And the biggest problem for this is if you look at where Claude and ChatGPT and all these guys want to go, they are building desktop apps that do precisely what this thing does.

24:04They're generic in that way. So I think it's more defensible, to your point, Mills, if this was specific to a particular industry and verticalized. but this is the this is the prototypical business that should be terrified uh based on what the core models of chat gpt and claude and these guys are doing because they want to own this space everybody needs this thing and they're going to build it and they're they're halfway there like if you use claude desktop or codex like from chat gpt like it does exactly this well it's kind of the shopify store problem right where you know so many people are using shopify for e-commerce and And then there's this whole secondary marketplace that emerges of Shopify apps.

24:44And my impression, and I'm not an e-commerce person at all, everybody knows that, is that Shopify kind of protected the masses, so to speak, by not competing with the Shopify app marketplace and just owning it, but not stealing, you know, those underlying apps and embedding them themselves. And that won't happen with something like this. They'll just go ahead and absorb it. And I think to Michael's point, so building generic kind of catch-all products within AI probably isn't something that lasts long-term. building products that use your domain expertise, something that you know specifically really well.

25:29I think that's where it's at. Like you could do something like this business model with something very specific that you know really well. I think that's where I would want to be. It's where I am. That's what I've been building, yeah. Well, Sam Altman had a line when was asked, he's the head of ChatGPT. when somebody asked him, these people building businesses on OpenAI, when should you be scared of OpenAI and when you should feel like you're in a safe place with OpenAI? And he gave an answer that I thought was perfect because it works for almost every single time you're building a business on somebody else's platform, which was, he said, look, if you as a partner of ours or you're building on our platform, if you are excited every time we make a new release, that means you're on the right track.

26:25If you are scared every time we make a new release that we might be running you out of business, that means you're on the wrong track. And the vertical thing that you're talking about is precisely on the right track, right? It's like, okay, ChatGPT is not going to specialize in roofing. Like, it's just not going to happen. But they sure as heck are going to create a general purpose assistant for business owners and executives, which is precisely where this guy's located in his business. One of the things that I've found a really helpful use case on is we purchase just tons and tons of materials, and they're all job-costed, and we get very specific quotes for each job.

27:06Even if it's the same material on a job today versus a job 30 days before or after, the quotes can be different. And then we usually pay for those materials, you know, many months later once they actually ship. But doing like a holistic review from like quote to, you know, invoice to payment, because these invoices, like I look at them every Friday and it's just like stacks, you know, of paper. and I don't like tying all those things together takes an incredible amount of time manually. But I like I was doing it this morning and I was thinking I'm shocked. I get all these cold outreaches about we can, you know, generate roofing leads for you or do you want to sell your business or can we help you with your, you know, phone system.

27:49But I have not gotten anyone that is, you know, something like this with a wrapper on it that is wrapped around invoice review or payroll review or things that like actually moved the needle for me significantly uh well i mean i would tell you like sit down with like codex or you know claw desktop and tell it your problem and then say tell me how to solve this uh treat it like you would like a smart chief of staff or an employee and then see what comes back um i bet you i bet you you would spend a couple hours on a weekend with a nice Coors Light or White Claw. White Claw. Michael, Nick Ultra.

28:31White Claw. And I bet by Monday you'd be like, cool, I've automated this whole thing. Like, I bet you it'll happen. So that's the power of all this is you don't have to know, like when I was growing up, to be a coder, you had to know like the language of the computer. Now the computer speaks our language, speaks human English. and it's just, it's changed everything. Everybody's a programmer now. Like that's the whole difference. I'll send you a picture when I do it, okay? Yes. Of the White Claw? Well, of me doing it with the White Claw. I'm partial to High Noons, but yes. Bartles and James. Oh gosh, you're going way better.

29:12Heather just had a flashback to college. Yeah, I certainly did. Cool, any other thoughts about this one? then I think we should go around the room and talk about thumbs up or thumbs down. I'm ready. I'm ready with my thumbs down. Go ahead, Heather. Tell us your thinking. Thumbs down. I don't think it has any staying power. I think you can build it for less than$500 ,000. Again, if you're in the market to buy something like this, you know enough already to probably build something pretty close to it. And I just don't think generic tools, they're going to be swallowed up by the apps themselves or the platforms themselves providing this service and getting better at it.

29:50So, you know, nice little business for someone to own for a little while, but not something you can buy for$500 ,000 and make money. Well, I'm thumbs down too, but I do think it's reasonably priced given that it's, you know, less than two times earnings. You know, every business owner thinks that their baby is beautiful. I would be really curious to hear this person's kind of counterpoint to the fact that this isn't differentiated and I just I'm very curious what they would say. I think it's a fair question to ask. It's a fascinating one because it's one of those businesses that for all the reasons Heather said, once you know a little bit you see all the problems with this business and also why it's so cheap.