I Went All In on AI. Here’s What I Learned.

In November 2022, I was running a profitable full-service agency. By March 2023, clients were asking if our work could be replaced by a chatbot. Both things were simultaneously true — and the gap between them was where I had to decide what kind of operator I was going to be.

The Wake-Up Call Nobody Wants

When ChatGPT launched, most agencies had the same reaction: nervous jokes, a few afternoon experiments, and a lot of “it can’t replicate real strategy.” My reaction was different. I pulled two of our account leads into a room and said: if this scales, our model has 18 months. I wasn’t panicking. I was doing math. The math wasn’t comfortable.

We had a team that was talented, expensive, and doing a lot of work that — I had to be honest with myself — was process-heavy and repeatable. Briefing cycles. Report generation. First-draft copy. Research compilation. These weren’t the things that made us good. They were the things that slowed us down.

The Decision: Go All In or Get Left Behind

I made a call a lot of agency owners weren’t willing to make: we were going to rebuild our workflows around AI before our clients forced us to. Not as a marketing angle. As a survival strategy. We started with a 30-day experiment — one process, one tool, documented results. If it saved us 10+ hours per week, we’d build a repeatable system around it. If it didn’t, we moved on.

What We Actually Changed (The Unglamorous Version)

Content briefing went from a 3-hour process to 45 minutes. Performance reports that took half a day to compile now took an hour. First-draft copy — the stuff clients would red-pen anyway — went to AI, and human talent moved to positioning, strategy, and the 20% that actually differentiates the work. We were transparent with our team about every change before we made it. Zero surprises. Every person who was affected got a conversation first.

We told clients too. That felt risky. It wasn’t. Two clients specifically said it made them trust us more. One referred us to a competitor because “you’re ahead of where we need to be.” Transparency in AI adoption, in my experience, builds credibility rather than eroding it.

The Results — and the Honest Caveats

The quarter after we completed the rebuild was our most profitable ever. We were doing more with the same headcount, margins improved, and the team was doing work that was more interesting to them. That’s the headline. The caveat: it took four months of uncomfortable transition, a couple of team members who didn’t want to adapt, and a lot of documentation work that nobody enjoyed. The outcome was worth it. The process was not easy.

What This Means If You’re an Operator

The question isn’t whether AI will change your business. It already is. The question is whether you’re leading that change or having it done to you. You don’t need a machine learning background. You need a hypothesis, a 30-day experiment, and someone willing to own the outcome. Start with your most repetitive, time-consuming workflow. Document what it involves. Test whether AI can do 80% of it in 20% of the time. If yes, build a repeatable system. Then move to the next one.

That’s what this site is about. Not theory. Not keynote slides. Operators helping operators figure out what actually works — built by someone who had real stakes in getting it right.