Meet the 30 AI Agents Running in the Background of a Remodeling CompanyWith Jeremy Wood of Neil Kelly Company

Eric Fortenberry: [00:00:00] Welcome to the podcast where we take a deep dive into the stories behind construction business leaders. We will share how they got started, how they found success, and the lessons learned along the way. I’m your host, Eric Fortenberry. Welcome to Builder Stories.

Eric Fortenberry: Welcome back, everybody. I am super excited about today’s episode. We have got a very special guest. We’ve got Jeremy Wood here. He is the senior director of IT for the Neil Kelly Company. They’re located in Portland, Oregon, but they have locations all across the Pacific Northwest, and actually one of the, uh, I think one of the largest, uh, design build remodelers, uh, in, in the country actually.

Eric Fortenberry: So really excited to have, uh, ha- have you on here today. Welcome to Builder Stories.

Jeremy Wood: Thanks, Eric. It’s a pleasure to be here.

Eric Fortenberry: Yeah. So, you know, again, I, I, I was telling you earlier, like, I’m excited because, you know, we, we got the IT perspective today. You know, normally I’ve got, you know, the builders and kind of the business leaders and, and, and people kind of focused [00:01:00] on, you know, everything, uh, business related, but I’m excited to geek out a little bit here.

Eric Fortenberry: We’re gonna talk about IT, about AI. I mean, it is a very, very, you know, rapidly evolving space. You know, Neil Kelly is, is, has been around for what, like 80 something years, I think? Is that, is that right?

Jeremy Wood: That’s correct. Yeah. It’s, uh, second, uh, almost third generation com- company now.

Eric Fortenberry: That’s awesome. You know, and, and, and again, I, I, I rarely come across, you know, a, a remodeling company who, who has a dedicated IT leader.

Eric Fortenberry: So super excited to, to learn more about why Neil Kelly, you know, decided to, to, to hire for IT and how you have been able to help push the company forward. But before we get into all that, why don’t you kind of give us, you know, a little bit of, of your background, you know, how’d you get into IT? What led you to, to construction, to remodeling?

Eric Fortenberry: Kind of give us, give us, give us a little, uh, little story on you.

Jeremy Wood: Yeah. So, uh, I was raised in California in Silicon Valley. My mom was a nurse. Uh, my father designed disk drives at IBM. And I always, uh, I wanted to do [00:02:00] something with computers that help people. Uh, so taking my mom’s, uh, nursing and her care medical background and then my dad’s, uh, engineering background.

Jeremy Wood: And so I spent most of my career in nonprofit healthcare. And, um, what I learned was you never, you never invent something that someone else has already done. And so when I worked in nonprofit healthcare, I saw a lot of companies were doing the same thing and had the idea, um, of essentially building shared data centers or shared systems and, and had a lot of success doing that.

Jeremy Wood: So I did it statewide and then nationwide. Um, and there’s a lot of, I think, benefits, economies of scale, standardization, and certainly cost reduction, right, that you can get by standardizing systems. So I had great success doing that. Um, sold off some of those [00:03:00] companies and then pivoted in or, or one of my best friends was a CFO of a public company that did utility construction.

Jeremy Wood: Um, and I like things in the real world. I love to learn. And so ran, uh, their IT shop and essentially built a nationwide shared services IT department. And so our strategy there was really to, um, either build or start companies internally or acquire through M&A. And what I would do as a public company, so we had a bunch of compliance, SOX compliance, uh, that we had to deal with.

Jeremy Wood: And so when we acquired a company, what we would do is we’d assess the MSP that they had or the, um, the technology, the internal people, and we would either, if it wasn’t good, we would do the MSP services ourselves. And then if it was good, then we’d just layer on the compliance, um, with that. Unfortunately, that company was either gonna make us [00:04:00] rich or go bankrupt and went bankrupt.

Jeremy Wood: So uh, when that happened, I, I really got a taste for construction and, and understanding and appreciation for, from it. And so, um, wanted to do more of that. And the opportunity came open at, uh, Neil Kelly and Dan here is a friend of a friend. And so I offered to, to kind of revamp the IT organization here.

Jeremy Wood: And Dan, our CEO’s focus has been, can we leverage technology to, um, I guess multiply the business, right? Improve the business. And so we have sophisticated systems, we have sophisticated people here, we have a data warehouse, and so we’ve spent a lot of time and effort, um, building out the data systems, the workflows, and the data analytics that, that help run the company.

Eric Fortenberry: That’s awesome. You know, I’m, I’m, I’m curious when, when you first walked in the door at Neil Kelly, like, what, what was the [00:05:00] first thing you noticed about how a construction company handles technology compared to, to your other experience in other industries?

Jeremy Wood: Yeah, I think construction’s a lot like healthcare where it’s certainly not cutting edge, right?

Jeremy Wood: And then, uh, there’s cost constraints as well. And so for me, the way I think about technology is it’s a pyramid. And so first you have to fix the dial tone. And so when someone picks up the phone or when someone turns on the computer, does it work, right? And so that’s where we started. Like let’s just get the dial tone fixed.

Jeremy Wood: And so there was what we call technical debt. There’s computers that didn’t work, servers that hadn’t been patched, um, firewalls that were not working correctly, right? And so we spent the first six months to a year, um, essentially fixing the dial tone, right? And then once you get the dial tone fixed, what you wanna do is you wanna start being the lever.

Jeremy Wood: And so that’s really on the application suite is where the rubber meets the [00:06:00] road there. So are the systems, the sales systems, the ERP systems, the accounting systems, are they meeting the business and working in an optimal way? And once you kinda get those dialed in, you wanna move into the top of the pyramid, which I call the driver.

Jeremy Wood: And the driver is really where IT’s helping drive the organization. And this is your analytics layer, right? And so can we do analyt- analytics around costing? One of the things that we have at Neil Kelly is, I don’t know, decades and decades worth of job costing, projects, other things. So we can leverage that data.

Jeremy Wood: So new, newer builders don’t have that rich information and we can look at, you know, the costs over time. We can look at our spend. We can look at where we can have cost advantages, right, or discounts with, um, with our subs or our vendors. And so that’s really where we want IT to be is a business [00:07:00] partner and driving some of the most critical decisions.

Jeremy Wood: So we spent the first part of the, the journey here at Neil Kelly fixing the dial tone or infrastructure, and then we started looking at the application layer and then the driver. Um, we have a sophisticated CRM here, and so we spend an awful lot of time on our sales process and on our, um, I guess our opportunities, our sales pipeline, and then the, uh, the forecasting.

Jeremy Wood: What we’ve all, we’ve, we’ve lacked is really that good construction ERP that looks at the, the backend of the process, right, which is project management, the selections, the customer portal. We’ve built, uh, quite a bit of that into, uh, into our CRM, but it’s a little bit kludgy, and so we were excited when we, uh, when we discovered your product.

Eric Fortenberry: No, that’s, that’s awesome. And l- love to dive into that some more. One, one question I had before though, like, you know, again, with, with, with Neil Kelly being, you know, older, you know, more [00:08:00] established, you know, more influential company, like, i- i- is there, you know, th- th - I’m sure a lot of people were, were very set in their ways.

Eric Fortenberry: They had, you know, very established ways of doing things. Like, did, did you run across, like, the resistance to change, you know, when you tried to kind of really get into those, those new applications, those drivers, you know, trying to, trying to implement and, and, and, and drive people to adopt new technology?

Eric Fortenberry: Like, what, what, what sort of resistance did you get and, and how did you, how did you handle that?

Jeremy Wood: I think that’s in any company I’ve been in, right? People, people have been doing things for a long time. They’re experts at it. They’re resistant to change. Um, and so I think any technology play is really a change management play.

Jeremy Wood: For me, the technology is easy, and with AI, it’s getting even easier. What you really need to do, and what I try and do is write one-pagers, and I call it what’s in it for you, right? And so, if there’s any sort of change [00:09:00] within the organization, I wanna know, um, for me, personally, I wanna know what’s in it for me, how does it make my life better, right?

Jeremy Wood: And if you deploy technology for KPIs or metrics that are just for management, they’re gonna be hit with resistance, right? But for example, in the sales process, if you implement, you know, some sales, some sales tactics that you want people to do, but then you make it easy that there’s tasks and follow-ups that automatically land in my inbox, right, or in my views about, you know, this is what I have to do so it reduces some of that clutter in my brain of having to remember things, what’s in it for me is I don’t have to think as hard, right?

Jeremy Wood: I’m, I’m more ef - more effective, I can sell more, um, and it just makes it easier for me. And so I, I think to your, your point, Eric, yeah, you’re always gonna have resistance and you have to put yourself in the, the mindset of, or the, the shoes of the people that are doing the work of what’s in it for me and lead with that.

Eric Fortenberry: Yeah, I, I love that. [00:10:00] You know, I, I, I always tell people, like, it’s so important that you, you help make sure that everyone understands the why, but I think, like, you know, you’ve taken it a step further with, like, you know, what’s in it for me? I mean, like, yeah, like, what a great way to frame it and, and help them understand how is it going to benefit them and their day-to-day, and you’re so much more likely to get buy-in versus if you just kind of come in with the heavy fist and say, “Hey, here’s, here’s what you need to use, and here’s, you know, how you’re gonna do it, and we expect you to do this, this, this, and this.”

Eric Fortenberry: Like, you know, it’s, it’s, it’s one approach, but I think it’s always better to, like, get their buy-in and help them understand, you know, why and what it will actually do for them, because it’s a lot easier to, to have people kind of join along than, than to have to sort of force them and, and, and push them along.

Jeremy Wood: I think it’s a lot easier than it was a decade ago, and I think the main reason why is the smartphone. And so you’d be hard-pressed to find someone now that doesn’t use Google Maps, right, or Apple Maps to find out where they’re going, right? And so you’re already using technology. Uh, I’m, I’m sure there’s [00:11:00] some folks that are still using paper maps, but it’s, it’s gotta be few and far between, right?

Jeremy Wood: And so Google Maps makes your life a whole heck of a lot easier, right? And, um, I think the problem with enterprise software is it’s not Google Maps, right? It’s not a clean, buttery interface. It’s easy to use. It tells you what to do that you can, uh, not have to dive down three screens or Uber’s the same way or Uber Eats or DoorDash, right?

Jeremy Wood: Those are buttery applications that are fun to use that scratch the itch of what you need. And, um, so I think that’s, that’s software of the future, right? It’s gotta be easy to use. And the what’s in it for me is I don’t have to dig down into 10 screens and figure this out and go through 20 hours of training, right?

Jeremy Wood: I want it to be, be easy to use and, um, the, the most important features are a c - you know, a press or two away I, I’m with my finger.

Eric Fortenberry: Yeah. So I wanna talk about the, the role of AI at Neil Kelly, you know, as, as, as, as a leader, you know, of Neil Kelly’s AI strategy, the training [00:12:00] efforts, like, you know, you have a critical role there.

Eric Fortenberry: So what, what does an AI strategy mean at a remodeling company? And, you know, are, are, are you sel- selecting the tools? Like, are you creating the rules for how people, you know, use them? You know, how, how do you do kind of the, the, the training of it? Like what. Tell, tell us a little bit more about that. What does that, what does that mean and entail?

Jeremy Wood: Yeah. So I, I think the AI strategy starts with AI is here and it’s only gonna get better, right? And AI is not gonna take anyone’s job, but someone that uses AI really well might take your job, right? And so it, it behooves everyone to learn it. Um, and what it’s gonna do is it’s gonna take what makes people special, um, and human in their roles and magnify it.

Jeremy Wood: So instead of spending a bunch of time doing a bunch of busy work or taking, you know, text or pictures out of Word and putting links into Excel or whatever, it can get rid of that busy [00:13:00] work and really amplify your, your creativity, right, or, or the things that, that make you good at your job. So that being said, the strategy is AI is here and it’s only gonna get better.

Jeremy Wood: And the way that we think about it tactically is there’s AI at the edge and there’s AI at the core. And so at the edge, these are simple productivity hacks, for example, right? So, um, could be designers mocking things up in NanoBanana or ChatGPT, right, to help them with their, their renderings or their, their initial pitches, right?

Jeremy Wood: So that’s at the edge. It could be, I need to take all this, um, all this stuff out of this Word document and put it into Excel with hyperlinks to make a palm sheet, things like that. Um, uh, spell checking, right, checking for tone when you’re, you’re dealing with an angry customer. So those are all edge cases, and I think those are easy and people are, are gravitating to those well.

Jeremy Wood: And then there’s, [00:14:00] there’s the core. And these, this is when you start thinking about your business systems and your AI agents. And what we’ve done at, at Neil Kelly is we’ve created essentially an orchestration layer. And so these sophisticated frontier models allow you now to connect into your core business systems, so your accounting system, your CRM, your ERP, and you can hook those up through MCP or APIs and you can start interrogating them.

Jeremy Wood: And so you interrogate your data systems for what’s happening in the business, and then you augment that with standard operating procedures or other documentation that you have. And you can essentially create a model of what the business is doing from a data lens. And it is fascinating. One, you often discover that what people are doing in real life has drifted from the standard operating procedures and doesn’t match what’s happening in the data systems, right?

Jeremy Wood: And so there’s. And sometimes it’s [00:15:00] for the, the better and sometimes it’s for the worse. And so what would’ve taken a, a business analyst weeks of interviews and other things to do can be done in an afternoon. It’s mind-blowing. So that’s the, the first part of the AI strategy at the core is creating essentially a, a private model that is how the business works.

Jeremy Wood: And with that, we can start looking at business problems that we wanna fix or automations that we wanna develop. And so what we would’ve done in the past is we’d have a business analyst kinda figure out what business processes are broken and then maybe pass it over an engineer to either develop that automations and code or do it in Zapier.

Jeremy Wood: And so the, the, the foundation models or the frontier models can do that all for you, right? So you, you interrogate the data, you build out your models, and then you just ask simply like, “Give me 10 ideas of where you think that we could [00:16:00] improve or automate some of our most complex or dysfunctional business processes.”

Jeremy Wood: And you get a list. And so what I’ve done here is systematically gone, um, through those lists with leadership team and we’ve built automations or things in the background. Um, and so we have, uh, about 30 agents that run in the background here at Neil Kelly that do various tasks. So there’s quality tasks, right?

Jeremy Wood: So checks and balances of, um, looking at estimates, looking at, uh, scopes of work and making sure things are in alignment there. We’ve automated a bunch of our AP processing. Um, a lot of the reconciliation that was done between various vendors, all that’s been automated. We’re sending out, um, what I, I call it application in a box, but essentially it’s really nice HTML emails that have the right data for the right people at the right time.

Jeremy Wood: And so [00:17:00] I’ve worked with a lot of accountants, and accountants love spreadsheets, right? And when you’re trying to, to drive behavior change or certain key performance, uh, indicators or metrics, you’ll often send Excel spreadsheets to people and they have to figure out, like, what to do with it, right? And maybe there’s three things that are really, that you’re after.

Jeremy Wood: And so for a sales process, for example, maybe you’re trying to reduce the number of days to sale, right? And so there’s, there’s three or four things that really affect that. And so instead of s- sending a spreadsheet out that a manager that has to talk to their salesperson has to decipher, you can have AI do that for you, right?

Jeremy Wood: So you can send a pretty email that says, “Hey, this is what we’re trying to, to drive towards. Here’s the three metrics, and this is what you can do,” right? Or, or this is how they’re trending over time. And so taking all [00:18:00] that, I don’t know, complexity in the spreadsheet and having to figure it out, you can make a really nice email with some graphs that help people, right, and, and say it in plain language.

Jeremy Wood: And so the flywheel that we have here at Neil Kelly is that we have our most AI literate people on the edge doing, um, productivity hacks or, and I guess they’re not hacks, but just productivity gains or trying to figure things out. And if they have traction on them or if they’re working well, what we try and do is centralize it through our orchestration layer and then push it out to everybody.

Jeremy Wood: Um, and we’re having great success with that.

Eric Fortenberry: That’s awesome. I, I, I, I love hearing that. And, you know, I, I, we have the same exact, you know, some, some people are, you know, uh, kind of, I guess, more, uh, more excited about it. You know, they’re, they’re the early adopters, they’re pushing the edge, they’re trying to find different use cases, and so they, you know, often are kind of out there, you know, testing and, you know, [00:19:00] figuring out what works, what doesn’t.

Eric Fortenberry: And then they come back and they’re, “Hey, guys, you know, check this out.” And like, we’ll end up kind of letting them do, like, lunch and learns and present at Team Lunch where they can, like, share what they’ve, you know, what they’ve learned and what they’re finding success with. And then that just naturally gets everyone around them excited, and then they all wanna learn how to use it and start adopting those same strategies because, you know, they see the impact that it’s made on, on other people.

Jeremy Wood: I think that’s absolutely right. So we did a year. Uh, so AI has been out in earnest, what, about two years now, 18 months?

Eric Fortenberry: Yeah.

Jeremy Wood: It was last January that the, um, like, the coding agents and the workflows and co-work got really, really good that we could build this stuff. But the first year of our journey was really just using ChatGPT and helping people with that as a starter.

Jeremy Wood: And to your point, what we had is we had a lunch and learn, um, once a month that everyone came to, and it would be kind of three things. It would be, “Hey, here’s some new [00:20:00] tools that people have had success with.” And so Napkin AI, which does really nice presentations, right, was one of them. We’d have people demonstrate successes or wins that they’ve had.

Jeremy Wood: And so to your point, um, there was kind of a shared understanding, and then I’d always end the meeting with fun things. So nothing to do with work. And so, uh, a lady came and presented a fantastic vacation that she had had with pictures that she had planned, the itinerary with ChatGPT, right? We did a bunch of recipes or other kind of life improvements that we did.

Jeremy Wood: We had tools that people would use. And so we tried to keep it, um, you know, even if you don’t wanna use this for business right now, just start to use and think about AI. And I think that that, that went a lo - a long, long way with people. And so, um, I think that set the foundation here at Neil Kelly for AI journey, was kinda grounding people in, you know, these tools are here.

Jeremy Wood: They’re not scary. You [00:21:00] can use them for your personal life and you can probably make your, your day-to-day in the, in the office a little bit better too.

Eric Fortenberry: Yeah, absolutely. So earlier you, you know, you, you talked about, I mean, you guys obviously have, you know, very in-depth systems and, and, and, and things built out, but like, you know, I’m, I’m curious, like, how did AI help you determine or select job tread for your project management software and, and kinda what, you know, what, what led you down that path with us?

Jeremy Wood: Yeah, so I’ve been thinking about construction management ERP since I started here at Neil Kelly because it was a, uh, a part of the puzzle that we h - we didn’t have. And we thouht for a time that we were gonna build it in our CRM system. And the complexity, the cost, uh, it just didn’t make sense. And like I said before, you never in, in compute or technology, you never wanna [00:22:00] reinvent the wheel.

Jeremy Wood: You never wanna do something that someone else has already done and done better and cheaper. And so that being said, we went out to market countless times to look at various construction ERPs. We looked at the, the high-end stuff, um, and we looked at some of the lower end stuff. And we were always. It felt like we were always in the messy middle, and we also do handyman services as well here.

Jeremy Wood: And so, you know, there’s, um, Procore on one end, which is expensive, and you need a, a systems administrator and a team to, um, to manage it well. And then the other hand for handyman services, uh, a tool like ServiceTitan, you know, would probably be best to breed for us. You’d have to configure that. But neither one of those does the messy middle where we’re in where we do a little bit of everything.

Jeremy Wood: It’s mostly home, big home remodels, right? But we also do, um, thousands of handyman jobs a year. And so the fascinating part about [00:23:00] what we were able to do with AI, like I said before, is we have a model in AI of what this company does. And so the model has everyone’s job descriptions, it has all the standard operating procedures, it has our data in a, a private, um, uh, secure place, and then it has all the APIs and, and all the inner workings of how the system works.

Jeremy Wood: And so with that, we created essentially, um, requirements documents. And so we went with those requirements documents, and we have interviews that I did with people when I started about what was working in the business, what wasn’t working. We fit that into the model as well. And then with those requirements documents, went out and had more interviews with people.

Jeremy Wood: Are these right? Where could we tweak that? And then we had the AI models go out to market for us, right? And it’s essentially what you do in the old days as an IT director or an IT manager, [00:24:00] whatever, is you’d go out and you’d, um, you’d do, um, RFQs, right, or RFPs, you’d send it out to a bunch of vendors and you’d have a 20-page document and you’d get a 20-page document back and you’d go and you’d take it off.

Jeremy Wood: Like, that’s all done now. And I wrote a, I wrote a, a LinkedIn article not too long ago about how this went for us. And so what we did is we validated those requirements, and then we had the AI agents go out, a team of agents or a swarm of agents and go out and look at all the, the, um, the products on the market and then evaluate all the functionality and grade it.

Jeremy Wood: And there’s no product that’s gonna do everything for everybody. It’s just not the way the world works, right? And so we had a, we had a rubric of how we were gonna grade the different products and what was important to us. Um, and so estimating is super important to us, right, because it, um, leads to clean jobs, right, and good margins and company viability.

Jeremy Wood: Uh, user experience for [00:25:00] end users, super important for us. Like we talked about before, user adoption, ease of use is super important to us. These are construction workers. They’re not technologists at heart, so we don’t want something. So with that rubric, we got a list from the AI agents of the different products that we could look at.

Jeremy Wood: The thing that stood out most for me about your product is that you had open APIs and an MCP agent, and that you’re true SaaS. And so true SaaS for me is you can sign up for a month, you don’t like it, you get rid of it. To me, that, that instills the most confidence in me. One, it shows that you’re confident in your product, that you don’t have to wrap me up in contracts and then I gotta get out of it, that you’re just.

Jeremy Wood: It is what it is. If you like it, take it if it isn’t. And so I spent a couple hundred dollars and I set it up and I set up the APIs and I said, “Hey, you have the requirements here. Go build it for me.” And I almost fell out of my seat. In an afternoon, I had the, I had the system built and configured in a [00:26:00] way that I could demonstrate to other people that it would work.

Jeremy Wood: I called the CFO down here. She fell out of her seat. She’s like, “I can’t believe this.” I go, “I know. I can’t believe it either.” Uh, mind-blowing. And so, um, kudos to you for your business model of, uh, not locking everything down for, for being true SaaS, which is month to month. Um, what was so fascinating about this is, at least in my mind, in the CFO’s mind, we had made up our mind about what product that we wanted to use before I had talked to anybody in sales.

Jeremy Wood: Like, there was no sales call. There wasn’t salespeople lying to me about what the product did or didn’t do, which often happens, or always happens in Salesforce, or not Salesforce, uh, in, um, in software applications. And so, just a whole different paradigm, right? It l - it left us with the warm and fuzzies.

Eric Fortenberry: [00:27:00] That’s awesome. Yeah. I n - I noticed you didn’t. Yeah, you guys didn’t go through the sales process, didn’t, didn’t do a demo and, like, to just, you know, to have such, you know, a, a, an advanced, you know, large company make up, you know, and, and evaluate the product without even talking to us. I mean, I, uh, that was very impressive on, on, on our side.

Eric Fortenberry: You know, most people would have sent us some long RFP document and wanted us to kind of go through this whole, you know, song and dance and do demos and yada yada. But, like, I mean, at the end of the day, we’ve always believed, like, uh, let the product sell itself. Like, you know, it, it, it. The product is, is, it, it is what it is.

Eric Fortenberry: And, like, we, we think that it’s very important for people to be able to get their hands on it and truly try it because, you know, we, you know, a- any software company can do, do a, you know, a demo, but, like, are you really getting into the, the nuts and the bolts of how it actually works? Is it gonna work for you?

Eric Fortenberry: And, you know, it’s, it’s very easy for salespeople to just kinda, you know, skip through certain things and, you know, skirt over areas and, oh, yeah, it’ll do it, and kinda do some hand waving. [00:28:00] But, like, at the end of the day, like, you know, I, I genuinely believe, like, we wanna earn our customers, you know, and, and we want to earn, you know, the, the, that trust every single month.

Eric Fortenberry: And, and if the product doesn’t work for you, then yeah, like, you know, i- i- i - the, the, the, the whole, like, locking people into contracts and, like, you know, not even letting, letting people kind of, you know, try it, you know, is, is, is, in my opinion, it, yeah, it does, it does a disservice to all those other companies out there that aren’t transparent, you know, with their pricing.

Eric Fortenberry: And they force you, you know, to go through these demos and these, you know, processes where, you know, you kinda get through all, you know, it’s all about the contracting. It’s like, we don’t even know if the product’s gonna work or not, you know? And that’s, that’s why we offer that, you know, 30-day money back guarantee, like, get in there, try it out.

Eric Fortenberry: If it’s not a good fit, no harm, no foul, like, you know, cool. But, like, at the end of the day, like, you know, I just think it’s so neat that you were able to, you know, have very clearly defined requirements, and then you deployed these AI agents to go find you all the possible solutions and make that evaluation, and you gave them that rubric.

Eric Fortenberry: I mean, that, that, again, I, I, I feel like that is super [00:29:00] advanced and, uh, you know, r- really neat though to, to, to see how you deployed that.

Jeremy Wood: Yeah, I think the, the most fascinating part was, this was all in an afternoon, I had taken our job costing out of Intacct, Sage Intacct, and piped it into your system. And so for each, I, I put a handful of jobs in there.

Jeremy Wood: For each of those handful of jobs, it was getting up-to-date job costing. And then I did the same thing for the CRM. And so I made sure that the APIs that were connected, the agents did all this. And so by the time later that afternoon when the CFO came down, she saw real jobs, real customers, and job costings that reconciled to the penny in an afternoon, mind-blowing.

Jeremy Wood: And that, there, those were the, those were the pieces that I deemed to be the most complicated, and especially the job costing.

Eric Fortenberry: Sure.

Jeremy Wood: And the fact that we were able to reconcile it to [00:30:00] the penny in an afternoon, it, it just demonstrates, demonstrates the power of the AI.

Eric Fortenberry: You know, and the thought that like you didn’t even spend a bunch of time like going through any of the training, watching the training video.

Eric Fortenberry: I mean, you know, we, we provide just, you know, tons of, of training and help resources and our teams here, but like you just were able to figure it out on your own within an afternoon. I mean, hey, you know, my, my mind’s blowing here too.

Jeremy Wood: Yeah. Yeah. It was, um, it was amazing. I mean, this is, this is the way that software will be bought in the future.

Jeremy Wood: I, I also think in, what’s gonna happen in the future is that all companies will have an AI model of the way that the business works. And that as a salesperson or as if you’re selling software or something else, your agent will talk to my agent that knows and the agent will evaluate, “Hey, is this a good fit for us?

Jeremy Wood: Is there an ROI here?” And it’ll talk back to the other agent. Yeah, raise it to the humans to talk, right? Or don’t. But [00:31:00] I, I think that that’s, that’s kind of the way that these things are going. Yeah. Is that there’ll be an undercurrent of the AIs talking and trying to figure things out and then it’ll raise it to humans.

Jeremy Wood: Instead of the noise that you have right now, like I get 10 pieces of software, other, you know, things that might be helpful and maybe one in 100 will be, but I don’t have the time to sort through it, so I often miss things that might be good, right? And so I, it’d be nice if there was AI agents underlying that, that know, um, what’s going on.

Jeremy Wood: And we built a little, just a prototype of one that looks at all the things and then, “Hey, would this be helpful or not?” And, you know, there’s been a couple things that have been, uh, worth looking at.

Eric Fortenberry: Yeah. So can you, so, you know, again, we’re, we’re talking a lot about agents and, and, and, and I feel like that’s one of the more, you know, advanced concepts with AI.

Eric Fortenberry: Can you, can you kind of explain for, for everyone listening sort of what, what is an agent and, and how are you creating these agents and then deploying them to go, you know, do various [00:32:00] tasks?

Jeremy Wood: Yeah, absolutely. So you think of an agent as a, it’s just a program that runs on a timer, right? And the program does what you want it to do.

Jeremy Wood: So a very simple agent for us is that we wanna, we wanna have all the, all of the customers or lead, I guess it’s customers or opportunities in our CRM in job tread, right? So once, uh, twice a day it goes and it looks for any new, um, people that have shown up in the CRM and it syncs them across and puts their address and all the demographic information that we need.

Jeremy Wood: So that’s a really simple agent, right? That’s just a CRM sync agent. It’s the same thing with the job costing. So a couple times a day it goes into Intacct, looks for new, um, uh, AP or invoices that have come in and then it’ll sync that into Job Tread. And so it’s a program that runs on a, on a timer is what you think of, uh, about an agent.

Jeremy Wood: The, the [00:33:00] thing about agents is, and, and I guess the scary thing about AI and, and what we use AI maybe a l - not in the traditional sense of, uh, an AI agent. And so what you probably noticed with AI is when you go into ChatGPT or Claude and you type, you ask it a question, you don’t always get the same answer, right?

Jeremy Wood: But in business, you always want the same answer. Like I can’t have it making up job costs, right? That would be catastrophic or we can’t make up customers. And so what, what we have the AI do is we have it build the code that essentially does the data pipeline, that takes the customer from one system to another, takes the job cost from one system to another, compares, um, or pulls, you know, AP invoices out of one system, out of the Home Depot portal, out of here and does all the reconciliation.

Jeremy Wood: And so all that busy work and then surfaces it to an AP clerk to look at. But what we don’t do is we don’t, we don’t [00:34:00] have it probabilistically guess each time and give it, give false answers, right, or hallucinate. We wanna have it within guardrails. And so the program always runs at the same time, you know, if once a week or once a month or, or whatever, looks to do what it has to do based on the program that it’s running.

Jeremy Wood: And then sometimes we use AI to augment it a bit. So, so the example might be we’re pulling sales data or opportunity data out of the CRM. And so it’s the same program, it’s pulling the same data, we know it’s good, the number of open opportunities, how long they’ve been open. And then we might sprinkle a little bit of AI onto the end of it that says, “Hey, this has been open for this many days.

Jeremy Wood: You know, after 45 days, these things fall off a cliff. You know, these are three things you might try, you know, to do that.“ And so we’ll, we’ll, sometimes we’ll add a little bit of coaching from the AI to help, to help people. But really what we want is we want standardized data pipelines. And so the, the way that we’re using, [00:35:00] um, the Frontier Labs and their coding, um, programs is, ”Hey, take a look at the systems that we wanna connect or the activities we wanna do.

Jeremy Wood: Write a program, write a test harness to make sure it all works, package it up into a way that we can schedule it and then run it on a, um, on a continuous basis.” And it’s a little bit more complicated for us here at Neil Kelly because we have on-premise servers and data warehouses and we have stuff in the cloud.

Jeremy Wood: And so we have to have kind of the agent, uh, orchestration layer in our cloud, in our private cloud. If we were full API, like your product, and we were talking to someone else, we could do the AI agents in the cloud, which make, would make it a heck of a lot easier and more approachable for folks. So if you’re using QuickBooks online, and I know you guys already have a connect or something else online, um, you should be able to do it using the agents that are built into, um, ChatGPT or Cloud just by sending up the [00:36:00] NPC - NCP servers or APIs.

Eric Fortenberry: Yeah. So that’s, that’s interesting. So like it, you know, the, the, the agent, essentially what, what you described was it, it really, it wrote the code, you know, which e- essentially it’s, it’s following the same process every time to go and find, you know, the, the data in one system and, and put it in the other.

Eric Fortenberry: And so it’s, it’s not this, like, undeterministic, like, maybe I’ll go find something and maybe I’ll stick it in the right place over here. Like, it really is actually kind of writing out the exact process. So it’s, it’s, it’s technically using the open APIs and able to grab that data, but it’s done it in a way where, like, it created this, this program, it created this script that allows you to, to not have to go and, you know, set up, you know, your own infrastructure.

Eric Fortenberry: I mean, obviously, you guys have your own kind of, your own servers and stuff, but, like, for, for most people out there, they would be able to just have this as a program or, you know, a skill or whatever, you know, the, the different AIs call it - Yeah. That can run in that cloud [00:37:00] environment, but it’s not just going to the chat and saying, “Hey, you know, go, go find me this.”

Eric Fortenberry: You know, it’s giving it that very, you know, deterministic path so that you have that consistency because, you know, we’ve all heard and we all know that if you, if you go ask, you know, Quad, you know, to do something and you ask it the same thing, you know, 10 times, you may get different answers, you know, a few of those times.

Eric Fortenberry: And so, like, you, you have found that with these, these agents, these skills, they are, you know, a lot more reliable and deterministic because they are actually building that bridge instead of just figuring it out in real time each time. Is that, is that right?

Jeremy Wood: Yeah. And, and what’s so funny about this is this, this is all new to me.

Jeremy Wood: So I started in earnest developing these AI agents in January, and by March, I had developed a handful of them, and I got worried about it. Like, what am I building? Do I know how to support this? It’s all, it, and it, like, like, figuring out what software to use, it, it seems too good to be true, right? Like, what am I missing?

Jeremy Wood: How am I gonna shoot myself in the foot? And so I asked the AI one day, [00:38:00] I’m like, “Well, these AI agents that I’m building, like, how should I think about this? Should I be worried?” And it insulted me. So it, it came back and I wrote a b - a blog post about this. And, and it said, “You’re not writing AI agents, you’re writing data pipelines.”

Jeremy Wood: And it made me feel so much better, because data pipelines have been around for decades, and data pipelines just work. So I’m using the AI to build data pipelines that are deterministic that do the same thing every time, not really doing AI agents, but it’s something that a small company could never do before, right?

Jeremy Wood: Because you’d have to hire a business analyst, you’d have to hire a programmer, you’d have to have some infrastructure guys that know what they’re doing. So it’s what used to cost $100,000 of time and labor is an afternoon’s worth of work now. It is, it is mind-blowing. And I, I think what I’ve seen just in the last six months is that these tools were the, the domain of experts like me who understand the, the cross-section of business processes [00:39:00] and business analyst work and infrastructure and programming, but the tools are getting so much better that, you know, business users can use them and they’re only gonna continue to get better.

Jeremy Wood: And so, um, you know, positioning yourself like you guys have with the NPC server and your AI native, um, I guess product and thinking are gonna be able to enable people to, to use these technologies as they become more easy to use and more available.

Eric Fortenberry: Yeah. You know, I think it’s interesting too. I mean, again, you, you guys, extremely large company, you, you’ve got tons of resources, you know, but like you, you’ve, you, you already realize that, like, you know, you could have gone and tried to build something, you could have tried to incorporate it in something else, but like, you’re, you’re more strategically focused on creating these, these, these connections, these pipelines, these agents between existing systems that already have those, those frameworks in place.

Eric Fortenberry: You know, I see a lot of, you know, a lot of people who think, “Oh my gosh, like, I [00:40:00] can now go be a programmer and I can go use Claude and, and it can go build me this entire elaborate system,” you know, but I think what they don’t understand is a lot of, like, you know, the, the security is like, just because, like, you can, you know, tell Claude to go do something, like, there’s a whole lot more that goes into it than just being able to say, “I want you to go build this.”

Eric Fortenberry: You know, like, talk, talk to us a little bit about, like, what, what are some of those, you know, kind of h- how, how do you evaluate, you know, when is the right time to just try to go and, and, and create your own thing versus leveraging, you know, an existing proven system out there?

Jeremy Wood: Yeah, no, I think this is a, a key, a key point.

Jeremy Wood: So I’ve led development t - nationwide development teams. I’ve left, led offshore teams. Developing software’s hard, right? And keeping software running is hard, you know, you run a software company. Um, and so you’re managing everything from customers and customer requirements and unhappiness and, like, figuring out the roadmap and what you’re [00:41:00] gonna build all the way through the backend.

Jeremy Wood: And there’s software development life cycles, and there’s, you know, tried and true methodologies to make this happen. There’s, uh, development pipelines and testing pipelines. It’s, it’s exceedingly complicated. And so what the, the tools lull you into, or they abstract all that complexity, right? And so you can sit and you can vibe code an app that’s, “Hey, vibe, vibe code me up a construction ERP system.”

Jeremy Wood: And it’ll do a remarkable job. Like, it’ll create something that’s usable that you can use. And that’s, that’s what we talked about earlier in the conversation is those personal productivity gains, right? So on the, on the edge. And so people can spin this stuff up, and they can use it. And, um,

Jeremy Wood: and they can use it, but the problem is, like, how do you scale it, right? And then when it breaks, who’s gonna fix it? And so that’s, that’s the real problem here. And so what, the way that [00:42:00] we think about it is, yeah, you know, experiment on the edge, play with things, and then if it’s good, what we’ll do is we’ll incorporate that, you know, back into the orchestration layer, and then we’ll make it production ready.

Jeremy Wood: But to your point, the hard part is, like, keeping those servers up and running, right? And when there’s bugs or there’s, there’s always edge cases, right? Or, you know, data comes in or a, a field that it doesn’t recognize or whatever, and it can cause everything to, to blow up. And so, I think you don’t wanna get lulled into the fact that software is easy, even though it looks like it’s easy to build when you wanna scale and you wanna use it in production, you’re gonna have problems unless you have people that can support it.

Jeremy Wood: And what I’ve, what I’ve appreciated about Job Shred, but, but it’s also a concern of ours, and we can, we can explore this a little bit, is that I think you’ve really keyed o - in and done a good job on the core features that [00:43:00] make the product usable. Um, there’s some edge cases that we have when we went through our rubric that aren’t there, right, that we’d like, but the, the juice isn’t worth the squeeze, um, and that we might use AI agents on the, on the outsource to build those because they’re not in your product, right?

Jeremy Wood: Yeah. And there’s some workflow stuff or some other things that we can build on, but build on the margins. Um, and so that’s the way that I think about it to answer your question directly is, um, if someone else has already built the core functionality, it works well, and it’s at a reasonable cost that you’re not locked into a nasty contract, and that you can connect AI into it, it’d be foolish to build it yourself.

Jeremy Wood: If that product is missing some features or functionalities on the margin that you need, by all means, augment it with AI.

Eric Fortenberry: Yeah, absolutely. And, you know, I can’t tell you the number of times. I mean, I, I’ve, I’ve seen people who get sucked into this, like, kind of this, this sort of like a trap, right? Like, they, they, they say, “Hey, go and [00:44:00] replicate this and go build this.”

Eric Fortenberry: And like, it, it ends up, you know, where they find out that they have now just spent countless, you know, weeks and, and months. And like, it’s, it’s, it’s honestly distracted them from their core business to the point where, you know, they, they wake up one day and they realize they’ve blown through thousands of dollars of, of tokens and they, they have just kept going down this, this, this hole, uh, to, to be able to, you know, at the end of the day, realize that, man, may- maybe, maybe I shouldn’t do that and it wasn’t worth, you know, all of that time, you know, the, the, the, it’s, it’s, it is very hard for people when they don’t know what they don’t know, and they don’t understand that, like, AI makes it appear very simple and easy to kind of prototype something and get, you know, some sort of a solution that, you know, you think kind of works today, but it’s, it’s, it’s that much longer term, how do you scale it?

Eric Fortenberry: How do you maintain it? You know, is it secure? I mean, I, I’ve seen multiple people’s apps where, like, you just go into the browser and view inspect and bam, like, their API keys are exposed. They’re, you know, they, they all sorts of different things. Like, I, [00:45:00] I found one, one app that, that this lady, like, vibe coded and it, like, literally had an RSS feed that for no reason, it was publishing her entire customer data set, all of their names, their emails, their phones, like, all you had to do is inspect the browser.

Eric Fortenberry: I’m like, “Uh, I don’t think you intended on that.” And she was, like, mortified, you know? And she’s like, “How, how could that even happen?” I was like, “Well, you, you asked it to go build something and it said, How do I do this as quickly and efficiently as possible with, you know, total disregard to security and everything like that?”

Eric Fortenberry: And so it’s, it’s been a very interesting time to see, you know, a lot of people. And I think people are starting to kind of, you know, realize that it, it, it’s, it’s just because Quad can go kind of prototype something, it, it doesn’t necessarily mean that that’s the right, you know, production ready system for them.

Eric Fortenberry: And, you know, really, I, I, I like your approach of, like, why, you know, reinvent the wheel when you can find a solution that if, if there are a few things missing, you know, with that open API and with the MCP server and the ability to connect, like, you can add on, you can [00:46:00] bolt on that missing functionality, you know, for your, for your own, you know, specific company’s needs versus having to recreate the entire core and then get to those features anyways.

Eric Fortenberry: Like, it just, you know, I think a lot of times it does make sense when you can find companies that are very, you know, o- open API, you know, uh, AI-friendly, you know, it gives you so much more power and, you know, o- obviously, you know, you can have varying levels of cost. I mean, you know, you got, you know, the ServiceTitans and the Procores and, you know, they have very, very expensive solutions or, you know, there’s, there’s more, you know, people like us who are much more affordable.

Eric Fortenberry: Um, but yeah, it’s, it’s, it’s an interesting time, and, and I think people, you know, often don’t, don’t, don’t know and don’t realize how difficult it actually is to build and maintain and scale enterprise grade software that, you know, is secure and it is reliable, and you don’t have to worry about all those things.

Jeremy Wood: Yeah, I think you’re absolutely right. I mean, that’s, first and foremost for us is security. And so all of [00:47:00] the agents, the coding that we do is all on our servers, never exposed, right, all behind, um, hardened firewalls, uh, in the Microsoft data center. And so unless you have a security kind of IT, uh, infrastructure or DevOps background, right, you’re blind to, um, the mistakes that you talked about, right?

Jeremy Wood: And so you need to be exceedingly, um, exceedingly diligent or buy a piece of software that’s gonna do it for you. And I, I think what you’ve done with your APIs, right, is that you’ve, um, you’re able to get the data out in a structured way and a much safer way for the, uh, for the agents to, to manipulate and talk to.

Eric Fortenberry: Yeah, 100%. So, you know, obviously, y- you all have the advantage of, of having that IT infrastructure and, and, and the team and, and everything to manage that. What advice would you give to, you know, much smaller contractors, people who [00:48:00] are, who are earlier on, you know, they, they don’t have, you know, that, that, that IT, you know, infrastructure in place?

Eric Fortenberry: Like, where would you recommend that they start their AI journey? You know, are there, you know, are there specific tools that you’d recommend? Like, h- how do people sort of weigh the balance of, you know, using, you know, one of those frontier models, the kind of the, the, the security risk often comes up, and are they gonna use my data for training?

Eric Fortenberry: Is it gonna get exposed or leaked? Like, how do you kinda weigh that and, and s - where, where would you point people based on your knowledge and experience here?

Jeremy Wood: Yeah, no, I think it’s a, a good question. So we started the conversation as AI here, is here. It’s only ever gonna get better. And so it, it behooves us all to leverage it as a tool, like we do Google Maps or anything else.

Jeremy Wood: And so in your, your personal life or your business life, I think just having people experiment, um, and try and do simple things at first. So that’s, that’s kind of where I get started, and that’s, that’s the foundation. Um, the AI policies that I’ve [00:49:00] helped create over the years really dictate that if you’re putting sensitive information in that you need a company account, right?

Jeremy Wood: So we have company accounts set up in either ChatGPT or Cloud or the other providers. And if you get company accounts, Teams accounts, enterprise accounts, if you look at the terms of service, they are not training on your data. And so I think that’s the minimum requirement. So if you’re having your employees use it and they’re putting customer or sensitive company information in there, buy them accounts that are not expensive, and make sure that you’re getting the business accounts that have the terms of service that they’re not training on your data.

Jeremy Wood: You can also go into the settings and double check that they’re, that that’s turned off. And so that’s, that’s kind of where I’d start. I think the, the second piece is a lot of these tools now will connect into your mailbox, right, either Office 365 or Gmail. I would start with a read only, right? [00:50:00] And then you can have a simple agent that does something like, “Hey, summarize yesterday’s emails for me at 7:30 so that I have a new email with the summarization of the emails and what, and my to-do list for the day.”

Jeremy Wood: And so not ever writing anything, right? You always wanna start with read because you wanna make sure that the AI’s doing what you want it to do before you, um, um, before you commit anything. But I think simple agents like that, read my emails, you know, read my calendar, help me, um, help me get prepped for the day.

Jeremy Wood: Um, and then you can start with that, you know, getting a little bit more sophisticated. So, you know, hooking up the MCP server into your software and then start querying, you know, simple things like, you know, how many customers were added in the last day or how many jobs were started or, or things like that.

Jeremy Wood: But I, I think more than anything, it’s the experimentation. And then I often use it as a though partner. So I don’t know what I don’t know, right? And so you can [00:51:00] have the agents go out and, um, you know, the, the prompt is something like, “Help me think about, you know, how, how should I think about, um, how should I think about improving this process within the organization?

Jeremy Wood: Or what are the trade-offs if I do this versus that?” And it, it’s not often 100% correct, but it’s, it’s just like when I’m talking to you, you know, you, you bounce ideas off one another and you say something and then it, it helps me think about it even more, um, fine-tunely or I have a, an inspiration based on what you said.

Jeremy Wood: And so I think that’s another good, good way to use it as well.

Eric Fortenberry: Yeah. No, absolutely. And, you know, it’s, we kind of while back sort of went through that same, like, you know, we encourage our employees to have, you know, AI to use, you know, chat, Claude, you know, the, you know, kind of whatever their, their AI choices.

Eric Fortenberry: But like, we, we also had to say, “But hey, look, you know, if, if you want to be, you know, doing anything where, you know, you’ve got customer data or any work-related, you know, data or [00:52:00] anything like that has to be within our, you know, accounts, our, our, you know, kind of enterprise team accounts” because, like, it, it’s exactly what you said, like, we have that, that guarantee that, you know, that data is not being used for training, that it is gonna stay within our account only, you know, and I think it, it’s, it’s an interesting time though where there’s, you know, and again, I got, I got personal accounts, I got work accounts, like, remembering, like, which one you’re on and, and, and where you should do what, like, that is, that is where I think a lot of people sort of struggle is, like, you know, should I be paying for my employees’, you know, chat or Claude account?

Eric Fortenberry: And, you know, the, the, the answer is yes, if they are interacting with your systems and, and your client data, your information, because, like, you know, uh, uh, heaven forbid, they, you know, they leave one day, you know, you wanna know that they’re not walking away with all of that information in their personal account.

Eric Fortenberry: You know, when it is a work controlled account, you know, you immediately can disable their access, everything is staying right there, and it’s not leaving, and so that’s where I think people are gonna have to really be very mindful and, and, and really think [00:53:00] through and put in place these policies so that they, they can make sure that their employees are, are using it appropriately, and they’ve sort of separated this, this personal and, and, and, and work usage.

Jeremy Wood: Yeah, and I think those, those are good points. So th- this is probably part of your life as well, but I find that I spend probably 80% of my day in my coding tool now, right, in AI, and so it reads my emails for me, right? It drafts emails for me. It, um, it does the queries I want, and I can pull the reports I want.

Jeremy Wood: When I have to go into a piece of software and I have to click and hunt and peck, I get frustrated. And so the, the new computing paradigm really is AI is I talk to the computer because that’s the fastest way to get information in, and then I read out. And I think where compute is gonna be in a decade is that you’re gonna be talking to the computers, right?

Jeremy Wood: And I think that, [00:54:00] that you guys have, um, you guys are setting yourselves well up for that, that kind of paradigm shift. But I’m curious, um, so I will say this about JobTread. I looked at it years ago and I, I dismissed it as being too small for us. I, I gave it a little bit of a look, and part of it was, and, and I’m gonna give you great kudos here, is that the software is, it’s easy to use, it’s buttery, it, it looks simplistic, it’s got some complexity in it, but you have abstracted it from people.

Jeremy Wood: And so you have this, this like, “Oh, it’s for small, you know, construction people and anyone can use it,” but then you’ve bolted on this, like, AI stuff that I don’t know that many other people have. And so you get the best of both worlds. You get something that’s easy to implement and understand, but you get this, this richness of the, the MCP.

Jeremy Wood: And so tell me a little bit about, like, what was your mindset and, and why did you decide to, to pursue the AI path? [00:55:00]

Eric Fortenberry: Yeah, well, I, I mean, I, I, I have to, I give all credit to our, our CTO and, and my co-founder, um, you know, e- essentially from, from the very beginning, you know, he had the notion of, like, we have to be an API-first company, you know, uh, back when we started our first business, I don’t know, many years ago, it was like, you know, everything needed to be mobile first.

Eric Fortenberry: It had to be, you know, responsive and easy to use on any size device, you know, well, again, kind of fast-forward, you know, I think the, the approach, you know, when we started, and this was back in 2019, was, you know, we, we need to make it as easy as possible to be able to integrate, and for data to be able to get in and out of the system is, is just, it’s, it’s imperative in this day and age.

Eric Fortenberry: And so, you know, he had the notion of essentially, like, we’re gonna make the core of JobTread as our actual API. And so the, the UI that, you know, all, all, you know, all of our users, they log into and they use, like, that’s actually built on top of our own API. Yeah. And so, like, having that at the core, I mean, that is what enabled us to.

Eric Fortenberry: And, and, and, like, we, you know, we, we, we, we were kind of, you know, sort of [00:56:00] waiting and seeing for a while with the AI, like, kind of wanted to see how, how it would evolve. And then, you know, one day, we, we just, we realized, like, all right, you know, it’s at the point, you know, Anthropic had, had open source, you know, the, the ability to, to create, you know, these MCP connections.

Eric Fortenberry: And, you know, it, it only took us a few days to actually implement that because we already had that open AI, open API, you know, infrastructure in place that made it just immediately available to the, to the MCPs so that they could connect and integrate. And so, you know, we, we’ve always viewed us as, like, you know, our, our role here is we, we can’t do everything.

Eric Fortenberry: I, I don’t believe that there really is one software. There’s not, like, one magic bullet here that’s gonna be able to do everything. And when people try to do everything, like, you, you often can only be mediocre at best. And, and, and I think it’s important that, like, we recognize that we, we strive to be a platform that is able to create this connected ecosystem for companies where they can bring the specific solutions that they [00:57:00] need, and we make it easy to, to integrate and to play well with others.

Eric Fortenberry: Yeah. And I think that’s where we’ve really seen, you know, kind of the, the, the, the, the market and just with all the AI and with everything that’s evolved over the last few years, like, having that, that forethought, you know, at the very beginning is what enabled all of that. And, you know, I think now as, as we’re seeing, you know, more and more kind of, you know, coding and, and, and AI and all these things, building tools, like, we’re already there and, and able to be integrated and serving as that platform, but also recognizing that, like, we can’t go build everything under the sun.

Eric Fortenberry: Like, we can’t do everything. We have to, you know, enable others to create their own customization and their own solutions that they can connect and extend the platform where they wanna go. And so, you know, I think, I think that was just a, a very early on, you know, decision and, and, and a very important, uh, you know, key move that we made that, you know, I’ve, I’ve seen, you know, some of our other competitors, I, I won’t name names, but, you know, that they’re, they’ve, like, you know, at one point came out and said, like, “Oh, they were gonna [00:58:00] bolt on an API.”

Eric Fortenberry: It’s like, you know, after 15, 20 years, you’re not just bolting on an API. Like, you know, and I think they tried for, like, six months and they came back out and said, “Nevermind, but here’s a couple integrations we built for you.” Like, it, it truly is at the core, and, and, and I think that’s what enables, you know, the, this, this MCP.

Eric Fortenberry: And I, and I think, you know, the, the future here is AI and agents being able to leverage these existing, you know, solutions that, that can work well with them. And so they are these, these system of records. And so being able to build these, these data pipelines, these connectors, you know, it, it, it is, it is exactly what AI is gonna be so great at, is, is using the systems for you and doing it on your behalf at all times and not having to have people go manually, you know, click and type and do these things.

Eric Fortenberry: Like, that’s the beauty of where AI is gonna just, like, you know, thrive, is this, this automation, because it can create these very deterministic connections. Yeah. And so, you know, at, at the end of the day, you know, the, the whole thing is possible because of [00:59:00] our API.

Jeremy Wood: Yeah, that’s, uh, I think you’re absolutely right.

Jeremy Wood: The, the other thing that I, I will say that, um, uh, ingratiated Your product to me was the fact that it is so open. So, like, worse comes to worse, it doesn’t work out. I’m gonna just export everything I need via your APIs. Like, it’s your, your true SaaS, month to month, no contract, open APIs, integrate with AI, and then if I wanna leave, I can just leave.

Jeremy Wood: The data’s not locked up. No. It’s easy to get out. Um, so that confidence of, you know, the software is what it is. Take it or leave it. We think it’s gonna work well for you, but if it doesn’t, you know, by all means, you ha - you have access to your data. And so, um, good all the way around.

Eric Fortenberry: Awesome. Well, look, I, I, I’m, I’m, I’m so glad to hear that.

Eric Fortenberry: And there’s nothing that, you know, upsets me more than anything when I see companies that, that literally are holding their, their users and their customers’ data hostage. And, you know, [01:00:00] again, won’t name names, but I’ve seen several of our competitors taking away the ability to export data, making it harder, you know, for you to be able to, to migrate systems.

Eric Fortenberry: And again, I, you know, I, I, I firmly believe that, you know, we should treat our customers how we would wanna be treated. You know, if we’re not, if we are not living up to someone’s standard, if we’re not meeting your expectations, if we can’t deliver the solutions that you need to build and manage your business, then, you know, it, it, it is what it is.

Eric Fortenberry: You know, we, we want to earn, you know, our revenue. We wanna earn, you know, our, our reputation and earn the trust of our customers. And, you know, if, if it doesn’t work out, then, you know, uh, again, we, we, we don’t feel like you should have to be trapped into some contract and we should have to, you know, kind of force, you know, your hand on these things.

Eric Fortenberry: Like, you know, I, I think when you, you truly try to build the best product that you can, you listen to your customers, like, everybody can win here. And, you know, that’s why, I mean, we’re just, you know, ranked, you know, 155th fastest growing company in, in the country on the Inc. 5000. You know, Deloitte ranked us the number six fastest growing, you know, s- software, you know, IT company in, in, in North America.

Eric Fortenberry: [01:01:00] Like, it’s, it’s something that, you know, when, when you, when you build a great product and you focus on just letting the product sell itself, letting the product, you know, just being extremely transparent, making the pricing easy. I mean, we hadn’t even increased our pricing, like, over four years because, you know, we don’t need to.

Eric Fortenberry: Like, we, we believe that software should be fair and affordably priced and solar owner operators shouldn’t pay the same thing as a 50 or 100 or a thousand person company. And so, like, I think just all of that transparency all around is what has, you know, really helped us to, to continue just growing at, at, at the rate we have.

Eric Fortenberry: But, you know, it’s, it’s also having great customers like you who can give us, you know, feedback and help guide, you know, the things that we can improve at the core that will further, you know, enable you to have success with the platform. And I think we’ve done a really good job of, of listening to our customers and, and really trying to decipher what, what should we do versus, you know, what are some, some ways that they might be able to kind of create their own solution using our, our API or using AI.

Eric Fortenberry: And so, [01:02:00] you know, I think our team’s done a, a really good job of listening to our customers and taking that into account.

Jeremy Wood: So, so one thing I’d like to point out, I do some advisory work, um, for larger customers. And what I see in the, in the mid-size market, so 50 to $250 million companies is that the AI is, is being slow to be adopted.

Jeremy Wood: And it’s for a couple different reasons, but one is compliance, right? So you gotta think about security and compliance. Two is there’s not clear ownership in the company of who’s doing it. There’s political and turf battles. And so I think that there’s a real opportunity for small companies to have an outsides advantage right now.

Jeremy Wood: And so, like we talked about before, like here at Neil Kelly, and there’s, like, if we wanna do something with AI, it’s a quick conversation with the leadership team or the person in the, the division head, and we just make it happen, right? [01:03:00] Um, it’s secure, we know it works, but it’s, it’s not turfy, right?

Jeremy Wood: There’s not questions about security because it’s all buttoned up. And so I had the opportunity to talk to your CTO about, about the product, um, after we, we got it set up here. And I was impressed with kind of the mindset that you had there. And he said, you know, we, we listen to customers, it’s a small team, we’re nimble, and I think you have an outsides advantage there as well, right?

Jeremy Wood: There’s not a lot of, uh, compliance or politics. You’re just making the product good and better for people. Um, no giant roadmaps or, you know. And so I think I appreciate that as well as, um, you know, it’s gonna be a partnership and, you know, we’re pretty technically savvy here and we’re on the cutting edge of this.

Jeremy Wood: And I’m sure if there’s something that we needed or wanted that would benefit other people in the convers - or in your community, right, or your customer base, that we could just have that conversation. And if it made sense for both of us, we could, we could figure something out. So [01:04:00] I, um, I appreciated that as well as your, your nimbleness and your, um, I guess your, just your appetite to explore these new technologies.

Eric Fortenberry: Absolutely. Well, look, Jeremy, I, I, I appreciate you coming on and sharing a little bit, you know, behind the scenes how, how you all at, at Neil Kelly are, are thinking about AI, are implementing the systems, setting, you know, you guys up to continue growing and scaling. You know, obviously, you’re, you’re huge role model for, for all the other design build companies out there.

Eric Fortenberry: You, you do phenomenal work, but, you know, I’m just, I, I appreciate you coming on and sharing all this and giving some advice for all of those listeners out there who, who may be interested or earlier on in their AI journey, because again, the, the, the, the opportunity here is, is tremendous, the potential is, is huge, but it does take kind of finding, you know, those, those specific use cases that they can kind of go and start with.

Eric Fortenberry: And you don’t gotta try to do everything at once, just like really kind of dial it in and say, “What can AI do for me? Where can I leverage technology to build a more efficient process that will help us to scale and grow?” And, [01:05:00] you know, I think you guys are, are leading the charge and, and, and really great, you know, representations of that.

Eric Fortenberry: So thank you for, for sharing all of this and, uh, yeah, I, I’m excited to see what you all keep, keep building and, and can do a job tread, because I think it’s gonna be, uh, you know, very, very exciting future for, for everyone here.

Jeremy Wood: Yeah, it was a pleasure, Eric, and I look forward to our partnership.

Eric Fortenberry: Awesome.

Eric Fortenberry: Thanks. Have a good one.

Jeremy Wood: You too.

Eric Fortenberry: Thanks for joining us for this episode of Builder Stories. We hope you enjoyed the conversation and gained valuable insights that can help you in your journey along the way. Don’t forget to subscribe to the show and leave us a review. And as always, if you or someone you know has a story to share, please contact us at builderstories.com.

Eric Fortenberry: We’d love to hear from you. I’m Eric Fortenberry, and remember, every builder has a unique story. Keep building yours.