AI didn’t enter my business right away. It showed up earlier, before there was any formal use case, when I was just experimenting with it personally. This was shortly after ChatGPT came out. At that point, I wasn’t thinking about operations, productivity, or scale. I was simply curious. I spent time testing prompts, seeing what it could do, pushing and pulling at its limits to understand what kind of tool it actually was. It was interesting, but abstract. It felt more like something to explore than something to trust with real work or real decisions.

That changed about a year later, and it didn’t happen because of some sweeping strategic realization. It happened because I was bored, frustrated, and a little burned out. I was in the middle of rewriting standard operating procedures—again. Anyone who runs a business knows that SOPs are necessary, but they’re also tedious. You know they matter, but the work is repetitive and mentally draining, and it’s easy to put off or rush through. I was searching online for templates, just trying to find a clean format I could copy and adapt, when it finally clicked that AI was already doing exactly what I needed. Every time I asked it to structure something, it produced a clear, organized framework that mirrored what I was trying to build manually.

At first, the applications were simple and internal. Turning blocks of text into checklists. Breaking vague ideas into steps. Cleaning up documentation so it was usable by people other than me. Nothing flashy. But even at that stage, something stood out. The AI consistently caught phases I had overlooked. Our previous SOPs usually focused on how to prepare for a task and how to execute it, but they often missed the follow-up—what needed to be checked, monitored, or revisited after the fact. The AI filled in those gaps automatically, and it forced me to confront a broader issue: most of us solve problems through a single lens, our own. That’s not a character flaw; it’s just human nature.

That realization shifted how I thought about AI. It wasn’t replacing my thinking, but it was acting like a second person in the room. A neutral perspective that didn’t get tired, defensive, or stuck in familiar patterns. As a leader, you’re supposed to guide people toward solutions rather than hand them answers. That’s easier said than done, especially with employees who are excellent at executing instructions but struggle to step into higher-level problem solving. I started using AI alongside my team as a support tool, a way to help them explore ideas, reframe problems, and push past mental roadblocks without the pressure of feeling judged.

The next major shift came when I ran into the limits of what I was willing to put into an open system. Financial data, internal metrics, company numbers—those aren’t things you casually paste into a public AI interface. As a CEO, you already rely on multiple perspectives: CPAs, bookkeepers, controllers. Each brings value, but each also represents a fixed lens that can grow stale over time. I wanted a way to analyze data with fresh eyes without compromising confidentiality. That’s what pushed me toward closed-source agents.

Agents were confusing at first. I remember wondering why everyone was suddenly talking about them when I was already “talking to” AI. What made them different became clear once I started using them. Agents allowed me to apply consistent logic, rhythm, and structure to recurring problems. They could ingest information, organize it, and return insights in a way that matched how I think—data-driven, pattern-oriented, focused on incremental improvement over time. We now run a small number of agents across financials, bidding, sales, and early job-costing work. One of the most effective uses has been spec book analysis. You feed in a bid package, and the agent produces the information you actually need. You still verify the output, but checking its work is dramatically faster than starting from scratch.

Implementation, however, is the part no one likes to talk about. Bringing AI into a business is double the work for a period of time. People still have to do their jobs the old way while also feeding the new system. That’s a hard sell. You’re asking for patience and trust in exchange for future efficiency. What changed things for us was proof. Once one department started seeing tangible gains, others followed. Momentum builds when people see that the workload doesn’t just shift—it eventually shrinks.

I’ve tried building my own agents, and while it’s possible, the pace of change makes it difficult to justify unless it’s core to your business. You can spend months developing something only to see the same functionality rolled into a major platform shortly after. AI moves faster than most operators can, which is why we work with specialists who do this full time.

I’m not scared of AI, but I am intimidated by it. That feels like a healthy response. Every transformative technology triggers concern, and some of those concerns are valid. At the same time, AI has freed up hours of my workday. Not to work more, but to think more. To analyze what we’re doing and plan what comes next.

For contractors who assume AI isn’t for their industry, I think the misunderstanding starts with how they approach it. They treat it like Google, ask one narrow question, and walk away disappointed. AI works best when you talk to it, give it context, and let it expand your thinking. After that, the best place to start is simple: identify the part of your job you hate the most. For me, it was delicate emails and SOP writing. Those tasks drained time and energy without creating leverage. AI removed that friction and allowed me to focus on leadership instead of repetition.

At the end of the day, AI doesn’t replace judgment or vision. But it does remove the barriers that often prevent leaders from exercising them well.