Roofers Are Using AI Wrong and It’s Costing Them $$$

AI is costing roofers more money than it's making them due to these mistakes. So in this interview with Ryan Fontaine, who has been an AI expert since the words AI were mumbled out of the first person's mouth, he is an AI developer with Citadel who has contracts all over the world working with Fortune 500 companies, government agencies, and has developed tools for the roofing industry like Roof Hawk and Storm Scout. And in this interview, we're gonna break down those mistakes. We'll teach you the 5 use cases for AI for most roofing companies and the role that AI should play in your roofing company in 2026. This is a longer interview, but buckle up. It's going to be great. Now let's jump into this interview. Enjoy. Ryan, roofers are spending boatloads of money on AI right now, and a lot of them are wasting money. I'd like to chat with you and find out what, like, what mistakes roofers are making that's costing— having AI cost them more money. Than it's making them. So where do we, where do we start? Where do we go from here?

Well, I think there's a lot of different ways we can go on it. There's a lot of different scenarios that are costing them quite a bit of money. I think the biggest one is there are a lot of good AI tools out there, but they're starting with, you know, shiny object syndrome that this, this AI is going to get me more sales or it's going to get me more calls, leads, whatever more inspections, whatever broad metric they're doing. But when I ask people, okay, well, what does that mean? What does that look like? They really struggle at times to go past the broad scope. And if you're only doing broad scope, you're most likely not going to be utilizing the tool very well. And you might even actually be going the other way where the tool becomes a liability, not an asset. So what I mean by that is, is let's say, Let's say you wanted to say, okay, I want to get more inspections. I want, I want to do that this year. Okay. Well, what does that actually look like? How many inspections did you do last year? How many inspections did you do per month? Because obviously storm months, if you're a storm state, it's going to be higher than, than non-storm months. And then obviously in the north, if you got snow-covered roofs, you're not doing inspections in January, February, you know, maybe even December, things like that, depending upon the weather. So what does that actually look like? So if you're doing 50 inspections in April, what does success look like? Is it if this tool gets you to 60, is that successful? Or do you think you would have gotten there organically without any AI? Is it— is 80 your goal? And going through and actually setting an actual KPI for what you decide the goal of the AI deployment is. But if you use just broad of I want more inspections or I want more sales or revenue or all of that, That's gonna be problematic because you don't know where that traceability and accountability's going to come from. It might actually be that you just got more storms that year, so you had to have more inspections. Well, that would've happened with or without AI. That, that was irrelevant. But if you have it where, say, okay, this year we got one storm in— I'm in DFW, so let's say you got one storm in DFW and we surged from the prior year of no storms from 35 inspections to 70. Okay. Now you have your no storm year and your storm year. Okay. What do both of those look like? Do you want to get 100 if you get a storm? Do you want to get to 50 if you don't? Set those and work backwards. And that I think is the biggest one. And it's not just roofing companies that do that. That's across the board, whether it's Fortune 100, Fortune 500, stuff outside this industry. That is a very common problem is that They, they view AI as a broad tool, and it can be, especially agentic AI can do a lot of things, but it still comes down to, are you doing it in a way that actually helps us, or are you adding friction? But if you don't know what success looks like, you're going to have no real good idea on defining, did it actually help us, or did it hurt us? That'd be my first recommendation.

Yeah. So just to summarize what I'm hearing. Having a goal in mind to measure what success looks like, to build towards that, as opposed to, as you put, which I love the way you put it, shiny object. There's this new thing the roofers on social media are talking about it. Therefore, I need to start doing it using X, Y, and Z tool that does this.

Correct.

Yeah.

Because you can, you can sign up for a ton of tools, but if you don't know what your goal is, you're not necessarily going to use the tool right. And then that gets to the second problem. Now you've signed up for 15 tools.

Mm-hmm.

And now do you have tools on top of tools that no one's using?

Yeah.

And you get basically, you know, 15 SaaS subscriptions and you end up having, you know, the equivalent of 8 Netflix accounts, 4 Hulu accounts, and 3 Disney accounts, and you only have 2 people living in the house.

Right.

And, and that's the big problem that, that roofers run into is that you do get shiny, shiny object syndrome and it is fine. It happens. But you need to monitor that. And that's where defining what value looks like. Now you can audit the tool and is it actually delivering? And if not, is it the AI's fault or is it our internal system's fault? Because there are two aspects to success. The AI tool can do a lot, but if you're not using it properly, it can't deliver the value that it should be delivering. But it's an internal system problem instead of an AI problem.

Yeah. And no roofer ever said, man, I'm so grateful my overhead went down.

That's right. That's, that's the thing. You do have to control costs at the end of the day, you know, with well-run roofing companies, they've got great margins, far better than, than I ever thought as, as never being a roofer. They do well with margins when they're managed right. But you have to, you have to control your costs and AI can become a very big cost if you're not careful.

Yeah, it absolutely can. You know, one thing I really appreciated when you came in to run a session for RSRA on AI, we had a really rich discussion on applying the theory of constraints. And for those who are not familiar with this, this is essentially identifying a bottleneck in the company. And the analogy I took was from the book called The Goal, which the author's name is on the tip of my tongue, but it's a novel. Business novel. I know it sounds funny, but a business novel about manufacturing specifically on the theory of constraints. And I love the analogy in the book. It says, imagine you've got a team of Boy Scouts marching in unison, but then this guy in the center, maybe he's the young overweight kid that really can't keep up. And all of a sudden the back half of the line is lagging. Well, now they're not going to get to their destination on time because of the one bottleneck, that weak link. And then they have to be like, okay, well, how do we, how do we fix it? We have to allocate, first we have to, uh, to identify the problem or diagnose the problem accurately, by the way, right? If you start trying to fix something up chain, it's not going to solve it. And if you fix it too far down chain, you're not fixing the root. So identifying the problem. Second thing is allocating all of the, all of the resources to that constraint. So they, they stop the group, they get everyone there, they take his pack off, they distribute the load from everybody, and then they start, you know, whistling the tune so everyone can march in order and, and pick up the pace so everyone's moving in the same direction. And then you rinse and repeat. So identify the problem, uh, allocate all resources to, to remove the bottleneck or the constraint, and then, uh, rinse and repeat essentially. So what advice would you have for a roofing company owner or manager who is you know, in this place of, hey, I know we need to be using AI, I just don't know where to start. What advice would you give someone to help them be as effective as possible while spending the least amount possible?

I would look for, for tasks that are highly repetitive. Um, I would look for, uh, something where how many times are you doing the exact same thing and the cadence doesn't necessarily change. Uh, for example, you have an incoming call and then the person's interested and you book an appointment that can be automated with agents that can be automated with AI paperwork on the backend. You can have an AI agent look at pricing, look at estimates, look at, you know, run an audit of, did we miss anything on this estimate? Did we miss anything on this work order? Did we miss anything on the, on the order to the shingle man or shingle warehouse? Distributor. You know, there's all sorts of things that you can do that are repetitive, tedious work. I would definitely say that that's an opportunity that's ripe for it, but you also have to do a cost-benefit analysis. You know, if it's, you know, $800 a week for a human to do it and it's $750 for AI to do it, maybe look somewhere else first because it's not necessarily a huge cost savings. And AI is really best when it saves on cost. And the repeatability of something wears down a human. It's not necessarily the best use of a human. And remember, AI is an augmentation tool right now more than anything else. It's not a firm replacement tool. And so if you have a person that may be great at sales, but they're spending a lot of time on their paperwork, that's kind of where you want to see what AI can do to help you is automate that paperwork so they can go sell more because that's their actual skill set. And that's— that would be my recommendation is find your bottlenecks with repetitive tasks. What are you doing all the time? What are you doing that, that you really just wish a computer could do? And that's typically where you'll identify some potentially low-hanging fruit.

Yeah. Yeah. Very well said. I remember I was running a workshop for our RSRA members. This was last year in Phoenix, and one of our members spent about $600,000 a year in marketing. He said 75% of his leads are, 75% are coming after hours or on the weekend when they weren't staffed. And I, you know, and he knew this was not ideal. And in my mind, you're like incinerating hundreds of thousands of dollars. And when I hear that, I was like, the very first thing, it doesn't matter. You could throw a bad AI booking agent, but getting some way to have speed to lead. So one of the, this would probably be a good transition. If you were to, to highlight like the 5 main areas that you see most commonly, uh, uh, most common among roofers to focus their energy there as a starting point. Now, disclaimer, by the way, as we get into this, what I'm bringing up could lead to people responding in shiny objects, right? Ryan said, or Adam said that I should be doing this for the day. This is not that advice.

First, I agree.

To you, my friend. Diagnose the problem and figure out what problem you, you need to solve. And Ryan, you challenged me, which I, which I really appreciated during our session in inside the RSRA community when I was like, hey, you know, diagnose the bottleneck. You said, but also the cost of that problem and arbitrage. Arbitrage meaning tackle in order of priority, the most expensive problems to the least expensive problems, because we can in business get this like, oh, well, it's fun to build things and create things and feel like you're working towards it. And then you just solved a problem that might only cost the company $5,000 a year. Meanwhile, you have a gaping flesh wound that's costing you $450,000 of ad spend. Like, maybe start there. So that being said, you need to diagnose your own problem, but where would you say would be the top maybe 5 areas that, that are most frequent to help encourage our folks tuning in to dial in their attention of where are high-leverage areas of AI for them to focus on?

Well, the several that top, that come to mind right off the bat, they are definitely going to be speed-to-lead tools. If you are advertising and someone calls you at 10 PM, not only should there be a booking agent, but there should be an ability to respond and assure that person. So one of the things that And again, I am not a roofer. I've never been one, but you know, when, when someone's calling a roofing company at 10:00 PM, it's typically an emergency. They're not calling you because they decided that that was the time they wanted to call. So I would actually take it a step further with any kind of voice agents, any kind of voice AI, or even a human call center to kind of understand the time difference. And this is where you get to understanding your customers. A 10:00 PM call should also have kind of a calming aspect of it, of when they're explaining that, you know, my roof is leaking. Okay, we can be there first thing in the morning. We have an appointment and we're, we're going to send somebody out at first thing. In the meantime, some, some tricks are going to be, you know, put, you know, towels on your flooring or whatever the tricks actually are. I'm terrible at that. I'm not a roof. But, you know, I'm sure you guys have, or, you know, roofers have some great ideas on how, you know, a homeowner can be, can be brought back from the panic that, that is likely calling. Now they call at, you know, 5:00 PM, probably not a panic. 6:00 PM, same thing. They call in the middle of the night. That's an idea that you, you may want to do. Cause that's still a differentiator. Even if everyone was using voice AIs and everyone was using schedulers, that's still a differentiator.

Yeah.

So at the end of the day, no matter what AI tools you choose out of the list we'll go over, you still have to— still do have your competitive advantage. What makes you different? How can you be different? How can this AI tool help you be more obviously different than everyone else? I would say the next best tool would be some sort of independent inspection system using AI. Yeah, and obviously Roof Hawk is one of them, but, you know, putting a drone up there instead of getting up on the roof.

Yeah.

You know, in highly, highly experienced homeowner markets, Texas, Oklahoma, Florida, things like that, it's a big differentiator because A, it's not just your opinion. The AI also agrees the roof is trashed or it's not. And that also can be good because now you can say, here's your report. And, uh, you know, you don't, you don't have any damage. So don't let anyone else try to think that your roof is totaled and make a claim that isn't valid. Um, so that's another one that you can do. Helps expand your sales team easily too. I would say the third one that comes to mind is, is really an audit agent. And, and what I mean by that is there's a lot of stuff in paperwork that gets caught months and months down the road, it seems, in my conversations with roofers where, uh, you know, oh, you know, we, we had to supplement this because we, we didn't have that exact, you know, gutter length right, or something along those lines. And again, I know you can't see everything on a roof ahead of time. Like if the decking is bad, you can't see that because the shingles are obviously over the decking. So there will always be some element of supplementing, but having, having an agent go through your estimates and say, hey, you know, this is what it's doing. You know, this is what we're missing. But the other aspect of that is it also can, can validate what your margins are. So if you might have a salesperson that's signing up 10, 20, 30 roofs a month and you think he's a rock star. But you're making 5% instead of a higher margin that you would want. So that salesperson, while getting a lot of revenue through the door, isn't necessarily actually helping your company grow. And so that's another aspect that an, an auto agent can do is kind of keep your sales team, uh, honest about what exactly they're, they're doing out in the field. Um, and then from there you get into some more field tools. You get into some canvassing AI tools, you get into Uh, you know, recording your guys, uh, talking and engaging in sales training, AI tools. There's, there's a lot of different tools you can, you can utilize. But again, it all gets back to, you got to identify your own pain points. You got, and then you got to set your own KPIs. What does success look like? Because all of those tools can be very impactful. We, we built several tools in those categories and we have people that have come back to us and said, I can't believe what a difference this has made. I mean, it literally got me 50 more roofs this year doing it this way. But they came through with a plan. And by the same token, we've had people going, eh, I mean, it didn't really make much of a difference. Well, okay, well, what did you do? Well, you know, we, we hit the go button and then kind of sat back and did nothing. Well, that's kind of the reason it did nothing. And, and because AI is not magic, it's mathematics. It's very good at crunching numbers. It's very good at visual stuff. It's very good at logic, but you still have to give it the right data and you still have to give it the right processes. And if you can't do those, it really doesn't matter how great the AI is. It's, it's not going to deliver the value that frankly it should.

Yeah.

Yeah.

Super solid point. So I think we hit, was that 3 or did I lose track?

I think that was 3. I think it was 3. And then I kind of went through the last 2 quick with the, the, uh, canvassing AI tools and, uh, uh, recording and, and, uh, sales. Training AI tools.

Thank you. The analysis is huge. You know, one that I'd like to throw in, and I'm going to be doing a whole other session on this with a couple other folks, is just using transcripts. I know you and I, we're going to be breaking that down at the data center here at an event for RSRE members coming up, which I'm super excited about, by the way. Like, very, very excited about. There's a lot of excitement in the community. We just announced the event on Friday.

So that's awesome.

Yeah, it'll be a blast. But I know one of the things we're going to be doing down there is using transcripts to help fuel marketing and sales analysis, being able to write your ad copy, produce your marketing material, make your social media posts, get you video ideas. Like there's so much that is underleveraged from that. And what's amazing to me is just how, how few roofing companies are, are simply just bare bones recording and transcribing conversations, whether it's over the phone or in person, like you mentioned, with canvassing tools or sales recording tools. Et cetera. So awesome.

And let's, let's expand upon that briefly, if you don't mind, because that's, that's a very powerful one. Because you have good data. And the reason you have good data is the transcripts. Your potential customers are telling you what their pain points are. They're telling you what they need. That's a great thing to go through. Because transcripts, first off, one, no human is going to read all of them. Then it doesn't matter how many hours you have in the day, you're not going to do it. And you're definitely not going to absorb all of it in a way that's going to be, you know, give you good analytics as to what's going on. But that's where AI exceeds. Go through 10,000 transcripts. It might take the AI 6, 8 hours to process all of it. It will, but it can handle all of it. And now you're going to come back with insights that your literal customers and leads that are calling you are telling you what their pain points are. Think about how much more you can attack your market when you analyze what the pain points are, because most likely they're not what you think they are. You may get a couple of the top 10, but I highly doubt anyone's going to go 10 for 10.

Yeah, dude, I don't know if I told you this or not. I analyzed— gosh, I can't remember the head— the count was like 64 or something like that. Companies' roofing reviews all over the country, coast to coast, to get variation of like retail only, storm only, etc. And I used AI to analyze the most common emotional words or phrases used to describe the experience for positive reviews and negative reviews. And it was, it was wild to me. In fact, on the positive reviews, and this is a spoiler, uh, we'll put a link, uh, up here at the top of this video for somewhere up around there, uh, to, for folks that might want to jump in and watch that video. But little spoiler alert, one of the most common words that people used in positive reviews was the word professional. And what's amazing is like, you can't really say I'm professional. It's a feeling you deliver. So. That's one of the rare exceptions to the rule I'm about to share, which is just use the language that your customers use to describe the service that you provide. And I'll never forget, I audited this one of our members in RSRA. We do these audits for folks where they like share their website or marketing stuff and I give feedback. And I looked at this one website and I just felt like the headline was just, it was lacking that like reach off the page emotional connection. And I'm sitting there saying like, man, you should, you should, you know, get into your reviews and, uh, let your customers write your headline by, by pulling out the exact words or phrases. And I scroll down on his website and as many RSRA members have, they have amazing reviews. One of them was this, that, that from a customer that wrote this, the roof work was so good, I added gratuity to the bill. And I remember it word for word and I was like, Ruth, that's your headline. Imagine that you land on the page and it's a quote from this homeowner. The roof work was so good, I added gratuity to the bill. Like you couldn't make that up. And like no one sells you better than your customers. So I love it. Awesome. Ryan, moving on. I'd love to wrap up with this one question. In 2026, what role does AI or should AI play in the growth of a roofing company?

Positive question. I would say in 2026, you're gonna start to see a lot more uptake of AI. Right now, um, and, and when we do the data center, I have the, the stats and everything. Most roofing companies aren't using AI, and if they are, they're just using large language models. Um, they're using, you know, ChatGPT or Claude, things like that. They're not actually using true AI tools or using generalized tools. And, and that's fine to an extent, but the, the real power lies in, in individual tools. But I would say to me, I think probably revenue generation is, is probably going to be where there's a lot of impact with AI. You're starting to see it do very well at, at giving you more opportunities, more bites at the apple, so to speak, with, with people that need roofing and other services. I think it's starting to mature to a point where You may differentiate that still if it's a human or an AI, but it's not going to bother you anymore. Where a few years ago people were, they were rebelling against chatbots. Let's face it, everyone was kind of over it. But now they've been refined so much where I think that you're going to see the revenue engine uptake, you know, or generation of AI start to become more mature and more broadly used and useful. I, you know, I'd probably say another one is, is probably you're going to see some more AI in operations, you know, day-to-day operations, paper check or paperwork checking, you know, was, was the final check received on an insurance claim? Did the homeowner close out their, their, their accounts? You know, all are the other, their invoices. Are there any outstanding invoices? And I know that, and, and that could be manual, and I will, I do tell everybody all the time, there's really nothing that AI does that a human can't do. It's just, it does it faster, never gets tired, and, and does it 24/7. Yeah. Um, so, you know, I would say that those are probably the two that would do it. Um, I, I, I still don't think you're gonna see AI in a broad sense of, uh, say making, uh, claim decisions on its own, either from the carrier or a contractor side where the two AIs talk to each other, they reach you reach an agreement and a check gets mailed. We're probably still a little ways from that. I've heard some rumblings that people think that's coming. I'll take that this may be a more 2030 thing because it will eventually get here, but it's going to be more in 2026. I would say those two are pretty solid bets for 2026.

I love it. Ryan, thank you for being here. Folks want to tap into you, your brilliance, or follow along. What's the best way to connect with Ryan or support one of the companies that you've been working on?

I would say if it's, if it's some of the companies we've got, StormScout, we've got Roof Hawk, and then of course our main AI company, that is my job. I'm AI developer. I, Citadel Analytics. Feel free to reach out, Adam. You can share my, my email and stuff in the YouTube video if you'd like, in the down below area. We're not a big YouTuber yet.

Yeah.

Just after this video. But yeah, I mean, I love AI. If you have questions, always reach out. I'll be happy to help. From an actual AI perspective and integration and all that fun stuff.

Awesome. Ryan, thanks for being here.

Thank you. I appreciate it. Thanks for having me.