How We're Using AI to Scale towards a $70M Home Service Business

Home service now is seen as a safe haven from AI.

What can't AI replace? Where does it stay out of?

I think it stays out of the truck. As I think about AI in the trades, I think that most companies are trying to do it wrong.

Ooh, hot take.

Getting rid of the human is not the thing that we should be aspiring to do.

The pendulum is swinging halfway back, it seems like.

And we were guilty of it too. We tried to go full AI in call center and it's like, oh, well, that actually wasn't the right decision.

Where are you guys using technology replacement here?

A lot of it we're using to like. Welcome back to Owned and Operated, a top 200 business and entrepreneurship podcast. I'm your host, John Wilson. We run a plumbing, HVAC, and electric company out of Ohio, and for fun, I talk to my friends on this show about how to build their home service company. Today, I'm joined by my good friend and. Frequent co-host Jack Carr. Mr. Jack Carr.

That's Mr. to you. Mr. Big Money over here.

A fun thing about Jack is that he shares a name with a famous author. And whenever I go through airports, I see— I see your name.

Yes. Yeah. It also makes my name ungoogleable.

Yeah. Yeah, that part is pretty— that part is pretty much more famous.

Yeah.

There's a much more famous John Wilson, too. He's like an HBO producer. So that, that's kind of funny.

That's what happens when you have just a generic—

Yeah, it's like John Smith. Yeah.

Yeah.

Um, yeah, dude, let's, let's rip it. Uh, what are we talking about today?

AI. Surprise, surprise. I feel like AI is just now permeating through everybody's business and it becomes one of these talking points because it, because it's ever evolving. Yeah.

Yeah.

So I think that today we hit on how we're actually utilizing AI and what it can and can't replace and what that looks like for our businesses currently.

Yeah. You know, kind of an interesting aside to this as we get started. We did an episode, I want to say it was back in like February, but my sense of time is warped. And what we talked about was there were these like 3 distinct periods of like big growth in home service. And one of them was like the emerging of technology serving home service, ServiceTitan, Google My Business. Angie's List LSA, and that's like the mid-20-teens. And then there's COVID, and that like pulled forward and created a lot of like net new demand. So that was a pretty like remarkable time period. And then the most recent, which is just starting to happen, is like home service now is seen As a safe haven from AI. And I was having dinner on Monday night with a partner from like, I don't know, like a top 3 VC fund. And that was their exact thesis, was they're starting to— like VC funds are starting to deploy like very real amounts of money into the trades because they're seeing it as like 2 things. One, it's durable and it's a safe haven from AI, but 2, they can drive AI into that business similar to Uber, I think is probably my best example. They took taxiing and they made it technology and they're trying to do something similar with home service.

Yeah. And so I think that's interesting as we start to go through this because there's portions of this business, right? If anybody was listening to you, you said it's a safe haven from AI, but they're driving AI into the business.

Well, it's a safe haven in that, like, and this is their driving thesis is no one is, like, a robot's not gonna go into a call center in the next 20 years. And I'm like, yeah, I mean, to me that makes sense. I don't know. But so I think they're seeing it as durable, whereas they've been investing in SaaS, they've been investing in all these other things that are now being very disrupted. So I think they like the idea of basically like EBITDA arbitrage is, is how I think they're thinking about it. Where— and you and I had this conversation about that landscaping company a couple weeks ago that sold for like an astronomical amount of money for the size of business it was. And the only thing I could think of was they're looking at different math. Like their math is they can automate their entire OpEx stack and EBITDA goes from half a million to $2 million or something like— because that's the only way that that math makes sense. So anyways, that was this guy's theory too, is that they can drive like arbitrage by automating more inside home service and like driving EBITDA. I talked to a ton of home service owners, and if you're anything like the $1 to $5 million shops that I know, you're probably getting hammered with AI pitches right now. Most of them sound great until they hit the real world and just completely fall apart. The one that we keep coming back to is Vocca. What stands out is they actually understand how contractor businesses operate. This isn't just another AI answering service. Vocca handles inbound calls, outbound follow-ups, texts, web leads, even dispatching, all in one system built specifically for service companies. If you're on ServiceTitan, then this is a big deal. Their integration is deep, So you're not duct taping together 5 different tools and hoping that nothing breaks during your busy season. I also respect that they're realistic about AI. When a call needs a human, they've got a 24/7 live transfer built in so nothing slips through the cracks and your customers don't get stuck in a bad experience or like an AI loop. Owners using Avoca are seeing hold times basically disappear and booking rates jump, some by 30% or more. If you want an AI partner that actually helps you book more jobs, without creating more chaos, this is worth checking out. Book a demo at the link below.

Yeah, we've talked about that a few times in terms of— right, there's a few places that you can push leverage. Yeah, you can do people leverage, you can do monetary leverage. And what they're focusing on is their big thesis, which isn't anything that's absolutely crazy or new, is technology leverage.

Right.

We've seen it actually over the past 6 years. The difference is the technology leverage isn't CRM. It's not phones to CRM. or fax machines to CRM. It is no AI to AI, which is incredibly interesting. And did you guys go over— was there any specifics of like where they're viewing that huge level of arbitrage to hit inside the HVAC industry?

I think they were trying to figure it out. Yeah, I think it was more like they—

Which is why they're talking to John Wilson, the man, the myth, the legend.

Yeah, yeah, yeah. Super. Let's, let's call a plumber from Akron and like really work through this complicated topic.

Exactly.

But yeah, I think, I think they were trying to figure out like, where can it happen? So their idea is it can happen. So let's put a lot of money behind it. And it's a lot like they're starting to— like they did it in property management. There was a few others. So like this specific fund has already done this a few times. with old-school businesses and playing this out. And anyways, it was kind of funny because when he told me this and he's walking through how he's thinking about it, I'm like, dude, I just did an episode on exactly how I think this is what the next 2 years is in February. Are you a listener? But yeah, it was kind of funny. It was kind of funny.

From a granular perspective though, where do you think that Um, I, I guess let's start with what can't AI replace? 'Cause I think that's the easier version.

Yeah.

Of like this, this conversation is where, where does it stay out of?

I think it stays outta the truck.

In what way? 'Cause I mean, that's a generalist statement.

Like, I don't think, I think—

It doesn't replace the technician, is that what you mean?

Yeah. I think that like, as we're a couple years into this now, as I think about AI in the trades, I think that most companies are trying to do it wrong. And I think that— Ooh, hot take. Yeah, hot take. I think that getting rid of the human is not the thing that we should be aspiring to do, I don't think. When I look at our 11,000 or however many Google reviews, yeah, people say Wilson, they're grateful for Wilson, but they're mainly grateful for that tech that came out to their house and did a great job and built rapport and made them happy. And I think that people get very worked up about this idea of— worked up as they're trying to build to displace labor. And I just think in home services specifically, labor is kind of the alpha where I don't think anyone should aspire to do that. So I think whoever has the best labor wins. So I think it stays out of the truck. I don't think robotics is something I'm super concerned about because I think that would not be like— you'd be focusing on the wrong line item, I think.

But is it not robotics? Is it the idea of rather than going robotics, it either goes in a different direction of DIY? Or it goes in a different direction of the labor is now no longer yours, which I think touches a few other things, right? So I formatted my graphics card on one of my computers the other day, which is wildly out of my normal scope of things that I could be able to do. And that would normally be a technician that you send your computer off to that then formats it, or someone logs into your computer through a, you know, remote software and like does it for you. But now that's something that the information aspect of it is gone because I can get the information aspect online.

Yeah.

I'm playing devil's advocate because I don't think this is the way either, right? Nobody— you can understand what a contactor or capacitor does, or you can understand, you know, the changing out of a board and how to do it. I still don't think that the vast majority of the general population is going to be changing out their own control boards on a furnace.

Yeah. I mean, I think if people want to DIY, they're going to DIY. And if they don't, they don't. I'm probably very capable of doing a tremendous amount of things. That doesn't mean I have the time or desire.

I think that's fair. I do think it's worth bringing up though, is that I do think that the DIY aspect is more palatable when you have the information now readily available.

Yeah, I can agree with that.

And then, the other direction that was curious to me is more so from the idea of Does AI replace the back office allowing that labor? Right. We have, right, currently there's really, really good technicians, but they're really, really bad business people. And so they have to go work for a large corporation or a small corporation just to be able to handle the business aspect. So does that get taken away by the ability for agents to run somebody's business while they go and actually do the labor portion?

Yeah, I mean, I think so. I think it's probably just like, um, a, a good way to think about it is, uh, like lights-out factories. It still requires people, it's just as you scale, you just don't need as many. So like accounting is a good example for us. Like we just automated— we had automated, like half-automated AP like a year or 2 ago, and then we switched over to Sage Intacct, and then now we're— now It was like very manual for the past 8 months, like very manual. And like we just re-automated it 2 weeks ago and it's like, okay, that's done. So I think as we continue to grow, I think like the way to think about it is how do you supercharge your existing team?

Yeah.

And an idea that came up is like, can you allow Like, hey, we have all this technology to record every call and do everything. Well, is there a way to surface data from those calls that would like provide for a better experience? And I think that that's a good way to think about it. Um, I, I think that the ability to run 24/7, uh, like demand capture, uh, with like— we use Avoca, but like I think that that provides a better customer experience, but I think everyone's so focused on replacing and they're less focused on how does the customer feel about this thing that I'm replacing it with? And we were guilty of it too. We tried to go full AI in call center and it's like, oh, well, that actually wasn't the right decision. We should have customer first, what do they want? Well, they want the ability to call 24/7. Great. So Avoca does that for us. They still want the ability if they have to or feel like they need to, to talk to a human. And that does that too. So I think whether it's the field or in the office, I think the ability to have human is good. I don't know. Well, I went through—

Sorry, I'm laughing at the phrasing of that. The ability to have human is good. Yeah.

Well, so I had an issue at my house the other day with my internet. And I called AT&T and I couldn't get ahold of anybody. And there was no one. My assumption is there is no one to get ahold of. And I was sitting there thinking about that as I was throwing my phone against the wall. And I was just like, wow, I'm glad that this isn't the service that we're providing. But 6 or 7 months ago, we kind of thought about it. And started testing and like, it was tough. Like I had an issue and I couldn't get it solved. I still have not gotten it solved like 2 months later because like I literally cannot get ahold of anybody to solve it. Uh, yeah.

No. So I, I agree though. I think that human first is still the, the alpha here. It's still the most important. Um, that's not getting replaced. People still want the ability to talk to humans until AI gets to a point where it is essentially Again, human.

Yeah.

Right. Which once that happens, I think that there's a whole nother slew of dominoes that fall at that point. But where— so where do you think then it gives us the edge? We kind of touched on it, is it gives us the edge when it comes into like content or I guess data, not content, vast amounts of data that we can record and then mine for.

Yeah.

Opportunity.

Well, and I think it reduces friction.

Incoming calls.

I think it reduces friction.

Mm-hmm.

I think it can supercharge your team. So something that we're thinking about right now is we have Indiana, Akron, and Nashville, and Dayton, Columbus, mid-Ohio is dead center. So something we're thinking about is, can we set up a distribution center in Dayton? And is that juice worth the squeeze? And then can we use So like, yes, no. And I think the answer is yes. And I think the juice is worth the squeeze. And then the second, like, follow-up question is, well, how do you maximize it? And the way I'm thinking about it is what if you could— you could either have a 10-person purchasing and logistics team, or you could have like a baller as hell dispatcher that has like a logistics dispatcher. And as sales come across, like from Nashville or Fort Wayne or Akron or whatever, it sort of funnels through an AI system. I think we could build it of just like, hey, I need this product this day at this time. And I think we could either do that with a lot of humans or with somebody really talented, but I think it reduces friction to scale.

No, that's a good way to think about it. I think that's a really good framework is it reduces friction to scale as a auxiliary function to the human element.

Yeah.

Right. So the ability for capacity planning, AI dispatch, AI allows your call takers to still have the human interaction while then just entering it in and having the friction between taking the call and the technician showing up, reducing the friction in between those 2 points. So, I mean, it's all customer experience, right? They don't see the in-between. I still think, though, Like, I know I'm backtracking here just to finish this point out. I am afraid, and that it was the other direction that we didn't get to finish the thought on. The other direction is, um, I am afraid of the ability though for the singular tech— essentially what happened with Uber was, right, technology came in, allowed the, the taxi cab companies to then become decentralized so that you're— you can run your own taxi without the larger company.

Yeah.

through a larger technology platform. And so that's what scares me is that AI will enable some version of that. Yes.

Well, I think that already kind of exists and I think it went past that. So if you think about what happened next, so that's what happened when Uber first came out. What happened next was taxi companies got onto Uber. So I don't know what percentage of Uber is just literally driven by a taxi, but it's kind of a lot. Like we took Uber in D.C. and I got in a yellow taxi.

Did you really?

Yeah.

I didn't know that. That's cool.

So like they use the product same as us using Angie's List or whatever. Like it's just a source of demand. And then the other thing that happened was there are minimum requirements and like someone's gonna uphold those. In this case, it was Uber. So like Uber's minimum requirement was like you need a taxi license, you need a car within a certain year. Okay. Well, some people couldn't afford that. So they still joined companies that just fed off of Uber. I was in New York for the first half of this week. Every single driver worked for a business. Every single one. Yeah. Uber, Lyft. Yeah. So I think it's sort of like it went one way and then it sort of bounced maybe halfway back.

The pendulum swings.

Yeah. But as I think about the trades, I'm a little bit less worried about that. Mainly just from like municipalities, like municipalities licensing, totally licensing requirements. Like, I don't know, I don't think that they're— because it's government and because it's so decentralized, like permitting and licensing sometimes are decentralized. Like in Pennsylvania, your state license— there is no state license. Your license is by the county. So like how many counties are in Pennsylvania? So I I think in order to like really drive something big, I don't know how they do it. I mean, and that doesn't mean it won't happen. It's just like you have to change an entire federal government structure, and I don't think that's going to. I don't I don't think that's likely because they tried briefly with the Doge Group or whatever, and like. That went to shit in a handbasket. Like, imagine dealing with like a billion Betty Sues in whatever county jurisdiction. And like, we still have municipalities that require us to literally drop off an envelope of cash to get a permit. So I don't know how you're supposed to like standardize that.

Fair. That is a fair point, is the government bureaucracy that we all hate on the daily is actually what's protecting us in the long run.

Yeah, it's literally moat. Because they they can't get out of their own way. They can't get out of their own way on it.

Remember that, folks. Next time you complain about your local government bureaucracy, they're protecting your business.

That's right. They're keeping you in business. You don't need more leads. You just need to stop losing the ones that are already calling, and that is exactly what Quo fixes. Quo brings every customer conversation, whether it's calls or texts or voicemails, into one shared thread that your entire team can see. So whoever jumps in Next has full context instantly. There's no more scrambling, no more crosswires, and nothing slipping through the cracks. It's gonna keep your team aligned, speeds up your response time, and makes your customer experience feel dialed in every single time. If you want tighter communications and to lose fewer leads, check out Quo at quo.com/owned. You'll get a 7-day free trial plus 20% off your first 6 months. Yeah, so I think some other areas that are kind of fun, That give, give like an edge. I think reporting has been really interesting, being able to digest like mass amounts of reporting. Price comparisons has been really wild, like being able to take like, hey, this invoice from this vendor, this invoice from this, like where can I get best price? And I think that that's another example of you can either have 5 humans or you can have 1 person with like a little GPT. Marketing's been really powerful. Like, sort of, we haven't automated our call board yet 'cause we're a little like scared of, you know, giving something that hallucinates like control over our daily budget. But you can do a lot of reporting automated. Like, we ended up, we created a dashboard inside our business that used to have a full-time person like just focusing on that all the time. And it was like daily leads and ROI by week. By channel. Very intensive. We're 5 locations, so like did it by 5 locations and now it's like automated. Yeah.

So I mean, it still is, it ironically still seems to revolve as much as people are trying to push it through to the individual side. It seems that it, most of the value that we've spoken about today revolves around data and data extrapolation.

Yeah.

Do you have anything outside of data, data extrapolation that you're using AI for?

I think there's some benefit to coaching, but this is back to human. I think when AI first came out, the Rillas of the world became very popular. But I also think my sense of that technology is it is a tool, but it quickly became— it was pitched as necessary. But then what happened was you removed the human. And sales are human. Sales is a human-to-human interaction. And I think that like recording and like giving feedback can be a very helpful tool, but like it pales in comparison to just doing a ride-along.

Really interesting. The pendulum is swinging halfway back, it seems like.

Yeah, I think it's exciting as it deploys. And then you're like, okay, here's the reality of it is do you want your sales leaders not doing ride-alongs? And the answer is like, of course you want them doing ride-alongs. Like that's a really important function. Yeah.

I mean, and it transcends— this is where it gets complicated, right? Because it transcends beyond just the data, right? A ride-along doesn't only signify, hey, my manager is going with me, he's working with me, versus, hey, there's this thing that listens to me and gives my manager automated feedback, then my manager might pass along to me. Like, there's a cultural aspect, there's a friendship aspect. There's just so many more intricacies to a human-to-human— interaction than just like, hey, you said this word 63% of the time. When you use this word, it reduces sales.

It's a tool. It's just like, yeah, I think—

It doesn't replace the person, which, I mean, I think a lot of individuals initially hoped it would.

And I think like, as we're— and I, I, I'm not trying to make this like an Avoca pitch, but I, I think like, I do like the human-in-the-loop component because I do think The idea of that is helpful, is like, hey, this is meant to augment and it's not meant to fully replace. And I think frankly, that's right. Like, I think that's the right move.

And so that's, that's in reference. We've been referencing a lot kind of the, the benefit or the, the downside is around the human interaction. But what are you guys looking at from a like a technology replacement because you said, hey, we replaced this ability to look at AP and compare AP to each other. Like, to me, that, that should be replacing a preexisting technological function like a ServiceTitan or a, like a CRM or something like that. Are you— where are you guys using technology replacement here?

A lot of it we're using to like ease up tasks. I also think So to give the listener some perspective, we finished a hair under $30 million last year. And this year, I don't actually know what we're going to end up at achieved revenue, but our run rate, because we have a few more partnerships we're about to do in the back half of the year, our run rate's going to be in the mid-70s. And so I think the state of our business currently, is like, this is a strategy that we're actively deploying, is, hey, our business, we won't triple this year, but like, we're 2.5 times. And how do we remove friction from the process? And how can we allow technology to help us keep up with that pace? And it's like, friction is a big part of it. Like, invoicing for Like, that's actually not a small problem. Hey, Wilson was getting 200 invoices a day from suppliers last year.

Mm-hmm.

Today, with 5 branches going on 8, like probably 600 invoices a day, like a 3 times volume of number of invoices. And like figuring out a technology solution to do that was important. Like it was an important piece of the puzzle because a lot can happen really quick if you're not staying on top of your supplier relationships and invoices. So like, if you just take that one problem, but like compound it across all the other things we talked about, well, like, hey, now we're managing 5 call boards.

Yeah.

Okay, well, how did technology help us do that? Uh, so I think like in the face of fast-paced growth, uh, we're less focused on like technology replacement, uh, 'cause we were last year. Like that was a big conversation for us was like, how do we replace existing tools with this? 'Cause it's less expensive. And right now, just because of the stage that we're in, we're much more focused on how can we add $10 million of revenue and it feel the same as it did yesterday, or even easier? Which AP, that's actually what happened. AP actually got significantly easier in the past month because we automated it than it was 2 months ago. And in that time, we've added a lot.

Were you using a human before to do all the AP, or were you using— yeah, so is there any situations though, because that, that's definitely a great utilization of, uh, AI is, is utilizing it to, to do those functions. But what about like technology replacement specifically?

Which is like I said, we haven't done much technology replacement this year.

Really?

Well, it's like we haven't had the time to think about it. It like, you're not like sitting there like Oh, I have a great one.

I know one you did. Fleet management. Like, talk about that.

That was last year.

Because I think—

Yeah, that was last year.

But I think that's a great example, right? Is like, hey, you could, you could have this, this SaaS, right? That for fleet, fleet maintenance and management, or we can just go and build an AI on Lovable in 15 minutes and save ourselves, you know, $30,000 a year.

Yeah, it was $20,000, $30,000 a year. Um, so, so yeah, a couple things happened last year which were kind of funny. I'll start with the why, but we basically stopped buying SaaS last year. We were just like, we're kind of done. And the theory was we can either build it or we should be evaluating more if we actually need it. There's some things you need, like you need Google Reviews, you need your CRM, you need the VoIP to answer calls. Answer calls, yeah. There's some things you don't have much of a choice on. Um, but I think like the industry has become very— technology's abundant and, uh, it, it's just so easy to drown in it and it's so easy to like, you know, continue to find new solutions, um, which I'm very— I'm pro like when it's the right move, but like sometimes it's just not the right move and it's just like a drag. Uh, so like fleet was one of them where like, hey, I can go to market and like get a fleet software. But it's an ever-scaling cost. Like that cost was going to be $7 a month per truck forever, which like, I don't want that. Like, that sounds ridiculous. So yeah, we built an alternative to it and that's been good. And I think what I—

Oh, I was just going to say, have you seen the graph of this kind of idea? It's like the graph in the last 6 months is like apps that have hit market has reached like a 300 or 400x. And then the apps that have been actually paid for in utilization has dropped drastically at the same rate because people are just going out and building their own. Like, I can build my own app now and I do not need to use this SaaS that had this very small kind of use case.

Yeah. Yeah. I think— yeah, I think that's probably our best example. We also did some— like, we custom built something for purchasing. We've custom built a ton of stuff at this point. We custom built our marketing dashboards. We've custom-built most of our stuff, but a lot of it's like, we stopped allowing SaaS to get bought. Like, SaaS had to go to me directly for contracts, and I just started like banhammering it.

Yeah.

It's just a real drag on like— and like, the business didn't change like for the worse. So I don't know. Like, even the AP automation, what we found out is like, we were already paying for that feature. In a software that we had. And it's like, well, we're paying like 10 grand a year for this feature. And like, we start digging into it and we're like, we're not even using it. And I think that's the other thing is like, okay, are you fully utilizing the stuff you're already paying for?

It sounds like if you were building today, like, how would you look at—

I think if I were building and not doing what we were doing to grow this year, if you weren't like doubling or tripling this year, I think that it makes a lot of sense to focus on technology replacement. I think if you are doing what we're doing this year, which I think that similar to that guy from that VC firm, I think that what we're doing is basically AI arbitrage right now, where we are buying businesses at a rapid pace for an independent owner, And we're deploying technology into them to reduce friction as we continue to scale. And I think that AI and M&A is becoming a pairing.

Yeah, which makes a lot of sense considering a lot of M&A outside of the actual operations itself is data and understanding the data structure behind it. Because you're culminating multiple assets from, you know, a 600-mile radius. You need that good data to make those decisions based on, based on location. Super interesting. I mean, I think if I was building today or like, and I was a— again, push me back 2 or 3 years, my focus right now would be to evaluate, like we've talked about in the past, evaluating growth or value, or like optimizing for growth or optimizing for EBITDA. I think that this year or next year probably would be an EBITDA optimization year where I would take my $3 million or my $2 million in revenue and I would try to optimize as much as possible to see how much more cash I could get out of the business. Because I think that there's a lot of opportunity right now to be able to get rid of some old technology that is costing you a lot of money.

Yeah, I mean, I think that that is good. I think it's a good idea.

Thanks, John.

Yeah, well, I think what— and this guy's idea from— and I agree, is like, I think if you can scale with less friction, like, can— the industry's gotten used to like 10 or 15% margins. Why can't 25 or 30% be real?

Yeah. And I don't think that those 2 things are mutually exclusive anymore, right? Historically, you would think you would see, again, like you said, scaling costs matches scaling business, you know, cost per seat, cost per truck. So as you grow, you have to pay more and more. I don't think that's any longer the case. So there's actually room now to scale and optimize, which is, which is an interesting kind of a newer And I mean, that was this guy's theory is that you can AI arbitrage basically.

And yeah, we're actively living it. It's pretty interesting. I think we're making some messes along the way, good and bad, but we are keeping up with it, which I think is interesting. And where all M&A, what we found out this year is where all M&A breaks, it used to be operations for us and now it is very much accounting. Like accounting is where it all breaks because we can onboard operations in like a day or 2. Like it takes almost—

Yeah.

I mean, not no effort, but like it's not that complicated. Um, but accounting is hard. Like it's really hard. Like we're— you and I are still working through stuff from accounting from like what, April? Like that's crazy. We're 90 days in. Yeah, it's just where it goes to die. It's accounting. So like whatever you can do to reduce friction there, I think is like very helpful.

Damn underwriters, man. It's all underwriters.

It's crazy.

It's crazy. Awesome. I think that's— I think we hit the nail on the head here.

If you like what you heard, give us a 5-star review. Tell my mom I'm popular on the internet. Thanks, guys.