For too many high-performance contractors, doing the best work often comes at the expense of running a healthy business. Custom diagnostics, complex quoting, and the paperwork burden of rebates and equipment registration eat into the margins that make quality work sustainable.
In this episode of The High Performance Edge, Peter Troast sits down with Furman Haynes, co-founder of WorkHero, to talk about how AI can change that trade-off. Furman draws a clear distinction between traditional software, automation, and true AI, and explains what that difference actually means for a small contracting operation.
He also walks through concrete, real-world examples of how AI is being used today to 1) automate job costing across disconnected systems, 2) generate custom, education-first quotes that help justify high-performance recommendations to homeowners, and 3) handle rebate processing without the manual legwork.
For contractors looking to improve their margins without sacrificing the quality of their work, this episode is worth a listen.
Episode Links:
Welcome to the High Performance Edge, the podcast for contractors who lead with building science and want to turn that advantage into a stronger, more successful business. Whether you're in HVAC, insulation, solar, or home performance, we dig into strategy, sales, marketing, operations, hiring, and customer service approaches that power stable, more profitable companies. Hear from top contractors and industry experts as we explore what it takes to thrive as a very different kind of contractor. Here's your host, founder and CEO of Energy Circle, Peter Trost. Greetings, everyone. Thanks for being here. Today, we're getting into something I lie awake thinking about, and I suspect a lot of you do too. How AI is actually going to change the day to day of running contracting businesses. There's an insane amount of AI hype out there, and cutting through the noise is going to be an ongoing topic on the podcast. My goal for the AI episodes is to be real and practical. My guest today is Furman Haines, cofounder of Work Hero, and one of the sharpest clearest thinkers I know on real world AI. What makes Furman interesting is that he didn't set out to build an AI company. He set out to solve the admin problem that is so challenging for small quality first HVAC shops and ended up in AI because that's where the leverage turned out to be. We get into the distinction between software, automation, and AI, it matters, and walk through real examples, job costing, custom quoting, rebate processing. But what I love about this conversation is where Furman lands at the end, that AI might finally start to close the gap between doing great work and the administrative burden of running a healthy business. This is a good one. Furman Hayes, welcome. Great to have you. Great to see you. And let's just start off with, you know, Puss Furman. Tell us about what Workhero is. And I think in particular, talk a little bit about how you became this de facto expert on AI and small contracting businesses. I don't know about that, but I'm really glad to be here, Peter. Thanks for having me. I'm Furman. I'm I'm one of the cofounders of Work Hero. We spend a lot of time thinking about the small craftsman in this industry. You and I both know some of these people. I'll give a shout out to those like Robin Henry at at Building Pinnacle out in Colorado or Caden Orin at Lewis Legacy. These are small HVAC contractors, craftsmen, who do really good quality work and, in some cases, don't have a full time dedicated office manager to run their office. And we all know the many, many contractors out there like that. And how Work Hero got started is is my cofounder and I just started spending a lot of time with these people. Ride alongs at the supply houses down in Florida at HVAC School Training Symposium. And it was really clear right away that the difference in many ways between a good and a great HVAC business comes down to the office management, the person that's making the invoicing and cash flow and customer experience decisions. And so we started by thinking about how to solve that problem manually or through expert expertise that we are we have hired onto our team at Work Hero and over time have have realized the opportunity around AI. But it started from this place of there's an admin problem in this industry, and we think that admin problem is actually existential to many small contractors. And if we can solve that, the rest should take care of itself. Cool. Interesting. Well, I think one of the things that I find particularly interesting is I know while you are relatively young and you started the company in the era of AI, you didn't intentionally launch it and nor do you promote yourselves as an AI company, and yet you're using it pretty regularly and and obviously very effectively. Tell us a little bit about that piece of how AI become became so embedded in what work is doing. So much of this comes down to how contractors use their existing systems of record. I'm thinking primarily of their field service management software like Housecall Pro, ServiceTitan, Jobber. There's a million out there. And these are powerful tools. Right? And these are systems of record that help organize information in the field and with customers. And what we saw again and again is, again, for that one, two, three truck operation that didn't yet have a full time person in house dedicated to managing their software, they weren't utilizing it effectively or or really fully. Right? They didn't have the price book built into it. They didn't have the, you know, the the invoicing workflows fully built out. Maybe maybe they weren't even scheduling out of it. I remember one of the first times I did a ride along with a guy up in New Hampshire. He was trying to use Jobber, but he was effectively using Jobber just like a glorified task management tool. He was, like, taking notes in it. But when I said, well, are you actually scheduling customers in? Are you invoicing customers? He's like, no. No. No. I just do that in QuickBooks. And so this is a long way of telling you that in seeing that these tools are underutilized, what we realized is that if we could hire expert office managers and offer them out on a fractional basis, we and and plug into their systems, We could become this expert human layer that sits on top of their system of record, and that's what Work Hero is. And then over time, we realized that there's a huge opportunity to not just be a human layer, but an AI layer that that sits on top of their system of record and takes action. And that action can look like, you know, a lot of back office work, billing, rebates, price books, registering the equipment. But it's again, it started from this place of seeing that these small contractors needed to use their existing tools better, and then realizing we could build not just a human layer, but, but an AI layer that, that can really help them improve their operations, save them time, improve their cash flow as well. Yeah. Yeah. Well, that's it seems like a really good segue to the you know, I think trying to get at the, you know, the challenge that so many people are facing, which is there's so much noise. Right? And there's so much talk about it. There's so much, like, you know, you're gonna go out of business if you don't sort of adopt here. Can you just set the you know, like, you just talked about software. Talk a little bit about the distinctions between software and AI. And and the other term that gets heard a fair amount here is automation, which, you know, which is a component, but necessarily some of the things you just described are automation that doesn't necessarily involve any AI. So so do you mind just sort of painting that picture for us? Yeah. So I think it's important to start with software, traditional software as we think of it, which is a tool. Right? It is a tool with a bunch of features, and it does what you tell it. So you have to be the one just like with any tool, you have to be the one clicking things or telling it what to do. AI is more like an assistant. It has context. It can make judgments. It can handle different exceptions or edge cases. And so that's the direction we're moving. That's when people talk about the the agentic era, if you will. That's all that it means is we're going from a tool that you have to manage to an assistant that's going to take actions on your behalf based on instructions or context that you give it. The idea of an automation is is sequence of steps that that send information or take action between different pieces of software or sources of data. And so automations can, as you alluded to, Peter, they don't have to necessarily be using AI. When we say AI, we primarily mean large language models and and their capabilities. Automations can be just code steps, deterministic code steps that just say, go execute on this. Don't make any judgment calls. Don't do any sort of the reasoning that that LLMs do. So, hopefully, that's I think tool versus assistant is probably just the the simplest at least that's the simplest metaphor that I have in my head that helps me understand Okay. Where we are Yeah. Where we are today. That makes a lot of sense. Would you paint the picture for us of you know, because you have this unique visibility into, you know, a lot of contractors that are at this sort of, you know, couple of trucks, you know, ten people kind of size. How would you describe the state of adoption? And I think maybe as part of that, answering that question, you know, what's happening with the FSMs and the other the other tools that people are using on a on a regular basis because all of them are incorporating AI pretty heavily? Yeah. So it's moving really quickly. Last year, I I believe service agents came out with a report. Last year, about fifteen percent of contractors reported that AI had implemented it in their business, and they had seen meaningful benefit. This year, that number is thirty eight percent. So we're seeing, you know, an increasing amount of adoption, but but that's still a minority of all contractors. So if you're sitting here today listening and you're like, I haven't done anything yet, you're not behind this this all this stuff is here, and and it's time to learn it, but you're certainly not behind. There is a gap between and we think about this a lot between the small and the large contractors in that adoption. We basically see a twenty percent delta between the small contractor and the large contractor in their current implementation and adoption. And that's that's what we obsess about and think about all the time at Work Hero. If you can't tell from my hat and everything that we think about, like, we wanna help the small independent businesses out there adopt, implement these tools, but we we wanna help you do it. Right? We wanna build custom automations for you. The the only other thing I would mention around you you asked about the FSMs and what are they doing today, and then what else is out there? There you can think of the the the the current field service management software as deploying a lot of AI already today. So when you think of the automated text that they're sending to the homeowner after they've invoiced a technician, that's been around for a long time. That's a version of an automation. Or the invoicing tools that they're releasing, which are helping it make it a little bit easier to invoice. These are automations that are that are built in. The really exciting point that maybe we'll get into more is we are now in what I would describe as the Cloud Code era. I know that ChatGeePTS Codex and there's other tools out there, but Cloud Code, which was only released about twelve months ago, has dramatically changed the ability to build custom software. And so FSMs are putting some AI into place, but really where you're gonna see, and you're already seeing so much action, is completely custom software for a specialized contractor in really narrow niches who are either building custom on top of their system of record, or they're building it from scratch. And this is, it's kinda where we are today, and this is it's the Wild Wild West, I think. Yeah. Sure is. The Energy Circle team will affirm the concept of the Wild West because I built a tool on my own for calculating the crossover point of when oil when oil heat prices go north of four dollars a gallon, when is a heat pump a more economical solution? And I did that all by my all by myself. Yeah. Thanks to Claude. Okay. There's this twenty percent gap between the small companies and the big company. Is that really entirely related to just sort of staffing issues and kind of the capability the the sort of resources that bigger companies have? That plus the I I would say there's the financial resources, and there's probably the most important resource, which is time. And that's the one that we think about a lot for the small business owner. But, yeah, if if you're a dedicate if you have a dedicated team of four CSRs and one accounts receivable person and one accounts payable person and two office managers, which is like a sort of classic shop that you're probably running, you know, ten to fifteen to twenty trucks at that point, Then you have you can just, you know, tell someone, hey. Go learn Cloud Code, or go learn the latest ServiceTitan module on, you know, on on Marketing Pro or Dispatch Pro, and implement all the latest AI tools. If you're your own small business owner and when at night or on the weekends, you're just making sure that every invoice gets sent out, you don't necessarily have the time to to learn these these tools. Yeah. Okay. Well, maybe let's let's go now to try to be, like, really specific and practical about some of the kind of applications. And I guess, is there is there one particular back office function that that you have is the biggest pain point for for contractors and where there has been a good AI solution? Yeah. There's a lot out there, but I think I would probably choose job costing because it's one where there's a lot of different sources of data. And what AI is best at is working with unstructured data, as they say. Basically so so let let's walk through it. So in in the case of job costing, after a job is complete, a technician, hopefully, records all their notes inside of their Housecall Pro. Right? And they they, you know, describe the capacitor or whatever else went went went on in that job. At that point, a few things need to happen. The first is there should have been a some sort of slip or invoice coming from the supply house that was emailed to the contractor. If that's the office manager, the owner, whoever's responsible for taking that invoice and matching the those equipment costs to the job. Right? For that to do for that to happen accurately, that often requires manual matching of of jobs, especially in the case where there are multiple jobs in a given location. Right? Because at that point, you need to really work off the purchase order number, not just the the address, which is often used as the unique identifier. So you have this information about equipment costs that's living in your email. Right? And you haven't matched it. You have the technician notes that are in your your field service management software. All of that needs to and then you have the number of hours that the technician spent on the job. That might live in your payroll software. That might live in a text thread that you send. That might not be anywhere at all. And so previously, I would say that I don't actually know the number here, but most small contractors I talk to don't have the time to effectively do this kind of job costing because it just takes so much time. Yeah. In the era of AI, what you know, when we talk about working with unstructured data, it can reference your email. It can literally reference your email on that email that you got from the supplier. It can reference a text thread or wherever the information exists about the number of hours spent on the job from the technician, and then it can also reference those technician notes inside of Housecall Pro. And so what you can do is you can generate the invoice to be sent to the customer, but then you can also match that invoice to your actual equipment costs and your actual labor costs and make sure all that gets its makes its way into QuickBooks. This is a process for for a given job. It could could take you thirty minutes or an hour to just chase down all that information and can be done in a minute with with the with the workflow and the right tool set up. Hopefully, gives hopefully, that's just one concrete example. I think that's super helpful and super practical. Do you it also sounds like the setup for it might be on somewhat on the on the complicated side. So is that something for Workhero clients that you guys do for companies? It is. Exactly. Our belief is that there's an really exciting opportunity to use AI tools to build custom automations. And the custom part, I think, is really important, especially for this audience. Right? You know, we're just coming off of an HPC where there's there's there's so much there's so many specialty and and high performance contractors out there who, through through you guys at Energy Circle and other friends in the industry, I've really gotten to know. And they might really care about, you know, a custom quote, right, that really doesn't just list equipment or, you know, numbers that they don't understand, but really educates the customer on, like, why should you use variable speed instead of single speed, or why should you think about, you know, the blower door test or whatever it is? And, you know, where Energy Circle talks all the time about educating the customer, I think that AI gives you the opportunity to take complex job information and build really custom quotes in this case that that that serve up the customer. So long way of saying, work here can do this for your customers. I see a lot of guys that are learning it themselves too, and that's giving me some that that's really exciting as well. Yeah. Yeah. It's interesting. You know, one of the things you're kind of inferring here is is a is a long standing problem that has existed for any companies that are diagnostically oriented. Right? They believe in the building science. They typically start a job by collecting a bunch of data. And one of the long standing challenges has been this idea that, well, I have to package that all together into an energy audit report. And there's been these, you know, huge time gaps from the time of the collection of the data to actually delivering it back to homeowners and, frankly, homeowners not even really wanting a long form thing. So it's really interesting that both there's all these new tools for gathering data. Right? The the LiDAR based, you know, dimensioning tools and automated manual Js and so forth. And so this is sort of an interesting way of expediting that whole process. Right? Is that is that happening? Yeah. That you know, I think of a a custom quoting tool that we're building for a a small contractor out in Colorado right now. Yeah. Where that's doing exactly a version of that. You know, one thing I'll mention about you mentioned data capture. We have the the really cool LiDAR technologies that are beak being able to accelerate the load calculation. We now have the ability to scrape information from photos. Right? And so the model and serial number and other information can get be extracted really easily. Hopefully, photos are either in their FSM software or in a Google Drive, and and we can pull that information out, which is important to register the equipment or the rebate. Then we then there is also voice, and I would say we're still at the very beginning of this across the trades, but a vision where a contractor leaves that leaves their their job, and instead of having to get home and type, type on their phone or type on their computer, you know, everything they remembered from that job, or while they're on the job typing in notes. You know, so much so many of us think much faster and more cogently by just speaking into our our phone. And Yeah. We are not far from a future where and and this is what a lot of the field service management software are trying to build right now too, a future where you can just talk into your phone for five minutes about all the different specs of a really custom job and everything that customer might need, and it can spit out a professional looking custom quote with your company's logo and brand and your preferred style. Really interesting. I mean, part of the picture you're painting, Furman, is is the ability you know, it's back to the Claude Code thing, like, the accessibility of developing your own applications, you know, is in some ways, in a very real world way, fulfilling what you hear a lot of, which is, you know, these software as a service is a threatened I'm curious. So if when someone engages you guys and they and you are part of the team, are these kinds of custom developments that you're describing part of the package? Exactly. So Yeah. When you work with Workhero, we first wanna learn your your process manually and do that work manually. So whether it's sending an invoice to a customer or or a rebate, we we do that work manually to start so that we can make sure that it's high fidelity. And then we we, at this point, have built, you know, a library of about two hundred automations already that are can be deployed on an account. But, usually, there's gonna be some tweaking because the actual requirements in a specific rebate program in a specific state might vary. And because, you know, with invoicing, contractors have different preferences for the way this information gets communicated, whether it gets sent over email or text, a lot of different preferences. And so we we take those preferences, tweak or tune the automation, and then deploy it and deploy it in the account. Yeah. That's that is really cool, and and it just speaks to what a great, you know, asset Workhero is in a company. You know, for those who may just be kind of at the beginning stages, they're listening, they're curious, they're, you know, like like, many business owners, including me, are laying awake at night, you know, trying to figure out how to move faster. Where would you recommend people start? Like, what's the, you know, what's the easy entry point here to just start building the muscle? Yeah. I saw this recommendation online, and and I've I've made it before to others, and I think it's a great one. Start in your personal life. Start in your personal life because everyone has something in their personal life that takes up a lot of their time and feels emotionally annoying to you. So that might be grocery planning or meal planning. That might be reviewing your email or something like that. Take one of those use cases and just have fun. Like, over a weekend, take two hours over a cup of coffee, and and I would recommend starting on Quad Code. There's even a more accessible version of Quad called Cowork that still hooks into your systems. You can look up Anthropic safety policies. They're they're among the best in in in the in the industry there. And start to make sure that all of your tools are hooked up appropriately, and so that's really important. Right? You know, Cloud makes it really easy through what they call connectors to make sure your Google Drive or whatever storage you're using, your email, everything is hooked up. And then just in plain text, and this is the important part, it is these large language modules models are increasingly good at picking up on what you need from very little information. My my biggest recommendation here would be to use their voice functionality to describe what you need. So let's just take the example of a meal planning tool. You could describe, you know, hey, Claude. Every Saturday, I go shopping for my family. I'm trying to plan out, you know, three dinners and five meal lunches and anything else. And I have these specifications, and one of my kids is allergic to this. And I want and I'm I want to be able to, every week, take a photo of my fridge and give you a couple ideas for what I want, and then you just give me the grocery list. And it will build this this what will effectively become your custom meal planning agent for you. Every Saturday, it will ping you and say, hey. I need that photo from the fridge. You can upload that photo, and then it'll give you the list. Something like this is is probably less stressful, hopefully, than your actual professional life, and we'll immediately show you the power of this. Once you have some confidence, you can start to, you know, tackle some some other Cool. I like that I like that example. Although, I'm not sure I want anybody, including a a robot AI seeing my refrigerator. What's in your fridge, VT? Let's wrap by just sort of fast forwarding a couple of years, and I'm I'd love this answer to be tailored to this unique niche of high performance contractors. Can you can you envision, like, what the picture looks like of a company that has kind of embraced, that is fully utilizing these tools, and really kind of running their business around it? You know, in a in a some reasonable period of time, the pace of change is so remarkably head spinning. I know this is a hard question, but where do you see the small contractor sort of looking like in two, three years? It's a great question. It's a hard question. I would love to to turn it back to you, and I'll just hear hear your answer. But I'll I'll say that, you know, for a long time, you and others have helped me understand that there's a there's a trade off that I think the small high performance contractor has to make, and that trade off is between profit and and good work, high performance work. And every single session I've attended at the symposium down in Florida, every single other conference or podcast I listen to talks about this trade off and how we handle it. And I am hopeful, although not completely convinced, but hopeful that there is a future where that trade off starts to change. I think for two reasons. One is that there's overhead in the form of custom quoting, in the form of follow-up visits, in the form of diagnostics that will be made easier or alleviated by by AI. And the second reason is because of customer education. I think you have an opportunity to do more with less, take information in your head, and translate it using AI into customer facing information that hopefully will be able to allow you to communicate your value. And so my hope is that in three, five, ten years, we'll be in a in a world where high performance contractors get to charge what they're worth and don't have to spend spend their nights and week fretting about the paperwork. Yeah. Well, I love that answer. I you know, one of the things that we're, you know, in all of our data, we're noticing, you know, changing very rapidly is how much better informed homeowners are as they come in in you know, come through the door for the first time. Now, you know, that's a double edged sword. Right? So they there's everybody thinks they're smarter than, you know, than the contractor because they've, you know, they've learned about building science on via AI. So, really interesting. Furman, how, what's the best way to get in touch with you or stay connected with you you particularly and and WorkHero? Yeah. You can email me directly. My email is Furman, f u r m a n, at WorkHero dot pro. We can put that in the show notes. And then you can find WorkHero online, w w w dot WorkHero dot pro, on YouTube, LinkedIn, all the other socials. So please get in touch if any of this resonates, if you're a contractor, if you just wanna talk about any of these these problems and issues. Great. Thanks for being here, and thanks for being such a clear thinker about this and a clear communicator about it. We all we all need it, and you're you're you're really generous to tell the world as as much as you have been doing in all these presentations and and here today. So thanks a lot. Thank you, Peter. The High Performance Edge podcast is brought to you by our company, Energy Circle, the marketing agency that for fifteen years has served the most successful high performance contractors in the country. Curious about whether we might be a good fit for your marketing and website needs? It begins with obligation, no hard sales discussion of your business and goals, and an honest assessment of whether we can make an outsized impact on your company. Check out energy circle dot com or email me directly at peter at energy circle dot com.
Subscribe to our newsletter to stay in the loop on new podcast episodes