Episode 1: Exploring AI and Data Ethics
Episode 1 of Talking Out Our Assets dives headfirst into one of the most important—and misunderstood—topics in the field service industry today: how artificial intelligence and data should actually be used.
Hosted by J.P. Cahalan, this episode features candid conversation with Jeff Prince, Chief Technology Officer at XOi; Greg Thoman, Principal Product Manager at XOi; and Denny Lawrence, Manager of Service Technology at Comfort Systems USA.
Together, they unpack the realities behind AI hype—discussing trust, ethics, and intention—why simply using AI isn’t enough, and how irresponsible adoption can lead to mediocrity, lost tribal knowledge, and erosion of critical thinking.
Rather than positioning AI as a replacement for technicians or experience, the group reinforces a core belief: AI should accelerate learning, reduce friction, and support better decisions—without removing the human from the loop. From diagnostics and data classification to ethics, bias, and long-term industry impact, this conversation explores what it means to build technology the right way.
This episode exists because the industry needs a more honest conversation—one that respects the complexity of assets, the intelligence of technicians, and the responsibility that comes with powerful technology.
Episode 1 Transcript: Exploring AI and Data Ethics
Greg (00:00)
And as we, as we kind of move toward the future, we start thinking about things like diagnostics. Well,
These are all tools that we would like to be able to leverage AI for, not just generative AI, but AI machine learning, all of that for, it requires a large data set of truth in order to be reliable. Because what we really don’t want to do is build a tool that is just based on the gut of a generative LLM.
And say, all right, we have diagnostics now. And it’s like, well, you don’t, right? Like you’ve raced toward mediocrity and you’ve put something out there. And the way that you actually build a strong tool for the future is marrying that generative capability with deep knowledge of subject matter. And I could talk about that forever.
J.P. Cahalan (01:07)
Welcome back to another episode of Talking Out Our Assets. I am JP Cahalan, your fearless leader and host for this endeavor. I am being joined by two of XOi’s best. I’m very excited for this conversation. They have absolutely no clue what I’m going to say.
The reason I’m excited about this conversation—I have Greg and Jeff from XOi, who are literally the backbone of designing and building our product. And when we think about new technologies and the considerations that need to be taken, this is such a green field space. There’s so much where we could run forward and do all these things, but is it the right thing and should we be doing it all?
Jeff Prince (02:09)
Awesome. Well, JP excited to be here. Excited to talk through this. Obviously AI is a big topic that everybody is talking about, so I’m excited to share how we think about it and how it’s changing what we do.
My name is Jeff Prince. I’ve been here about six years. My background is in software development, been doing that for about 25 years across billing, customer care, logistics, and referral management. I’m just excited to be part of this and to be impacting the folks in our world that keep it running.
Greg (03:00)
Hey everyone, thanks for having me JP. My name is Greg Thoman. I’ve been here about six years as well. I have a very technical background, but also a very mechanical and manual background. I come from all over the place on the product side, super into technology, but also into applying it to solve real problems that don’t have great answers. This is something I’m personally passionate about.
J.P. Cahalan (03:45)
I freaking love it. Let’s go. Let’s get excited.
So here’s where I start with kind of every conversation I have with anyone in regards to AI. How do you guys use AI today? What are the core elements that you’re using AI for in your day-to-day lives?
Jeff Prince (04:08)
Yeah, I’ll jump in first. It’s really interesting to think about the evolution over even the last year. This has changed a lot over the last three years, but especially the past year. It’s almost my default now. Whenever I start something new, I’ll brainstorm with ChatGPT or Claude.
The other place I use it is design and proof of concept. I’ll ask it to spit out a quick POC so I can iterate quickly. That allows me to go from idea to something refined that I can share with my team. We’re also using AI for code writing, code reviews, and boilerplate work—always with humans reviewing and understanding the output.
Greg (07:48)
I’ll echo a lot of that. Generative AI is very good at summarization, classification, and working in well-known spaces. On the product team, we use it to synthesize input from technicians, executives, calls, and meetings so we can ask better questions of that history instead of re-sifting it.
Personally, I use it similarly—but I still need to understand what it’s doing. It’s a shortcut, not a replacement for thinking. Another big use is classification—helping bucket large amounts of data so we can get to pattern recognition.
J.P. Cahalan (11:00)
You guys said a couple of key things—how you use AI and classification. One of the hardest things for people and businesses is accurately classifying events to get insights. That makes total sense.
If I’m a contractor listening, what should I know about XOi’s North Star with new technologies? How are we viewing this rapidly evolving future to support users the right way?
Greg (13:18)
When I think about positioning technology, it’s about accelerating learning and reducing effort—documentation, paperwork, analysis. AI is good at some things and bad at others. You can’t just point something at generative AI and expect it to be good. That’s a race to mediocrity.
When you pair known data with generative responses, that’s where power comes from. Diagnostics require large data sets of truth. We don’t want tools based on gut feelings of an LLM.
Jeff Prince (16:43)
The most important thing is starting with understanding users and their pain points. AI isn’t always the answer. Sometimes it’s just a screen.
When AI is the right solution, we maintain trust by keeping humans in the loop. Technicians are extremely knowledgeable. AI should provide suggestions, not decisions.
J.P. Cahalan (19:23)
I love that. How far away are we from guided operations where AI tells you exactly what to do?
Jeff Prince (20:49)
We’re a ways away. Generic AI lacks vertical knowledge. Our advantage is millions of completed jobs. We can narrow options to two or three and let technicians decide.
Greg (25:14)
When technicians aren’t turning a wrench, they’re searching, calling, asking questions. We’re essentially building the support engine—but with context captured so humans and systems can step in when needed.
J.P. Cahalan (31:46)
Let’s talk ethics. Are we losing critical thinking?
Greg (32:50)
Yes, it’s a valid concern. Outsourcing thinking erodes critical skills. Leaders pushing unvetted AI ideas down the org create bias and cognitive decay.
Jeff Prince (36:28)
There’s risk, but also optimism. AI shifts thinking to higher-level problems—but someone still needs to understand the foundations. Losing tribal knowledge is the real danger.
J.P. Cahalan (39:30)
Final fun question—ten horse-sized ducks or one duck-sized horse?
Greg (41:23)
One duck-sized horse.
Jeff Prince (41:28)
Same. Easier to feed. I’d monetize it.
Denny Lawrence (42:37)
My name is Denny Lawrence. I’m the manager of service technology for Comfort Systems USA. We’re a national HVAC, mechanical, electrical, and plumbing organization.
I use AI personally for coaching, practice plans, presentations, email summaries, and efficiency. If you’re not using it, you’re wasting time.
J.P. Cahalan (49:16)
AI won’t replace your job—but people using AI will. What’s your expectation of technology partners?
Denny Lawrence (51:40)
AI is table stakes. Partners should be bringing it to us. Not bleeding edge—but intentional and impactful.
J.P. Cahalan (1:00:13)
This is why I love these conversations.
Denny Lawrence (1:03:19)
Absolutely. Appreciate you. Thanks for having me.
