Don't Confuse Fun with Progress on AI

Time for an honest one. I lose hours in AI that move absolutely nothing in my business. I’ll surface from a session having built something clever, feeling productive, and when I ask myself what actually changed, the answer is: nothing a client, a lead, or my bottom line will ever notice. I teach this stuff for a living and I still fall into the hole. So this post is as much a note to self as advice.

Here’s the trap: AI is genuinely fun. It responds instantly, it always makes progress, it never tells you the idea is silly. Tinkering with it produces a feeling that is almost identical to the feeling of doing valuable work. Almost.

The test: name the number

Before any AI session, and honestly at any point during one, ask one question: which business number does this move? Leads in. Margin on jobs. Hours back in my week. Speed from enquiry to quote. If you can name the number, keep going. If you can’t, be honest that you’re playing. Playing is allowed. I like playing. Just don’t write it in the diary as work.

What moving a number actually looks like: a builder I work with, brand new to running ads, set up his own Google Ads with AI walking him through it, no agency, no consultant. Within a week it brought in an enquiry for a $250,000 renovation, exactly the job size he wanted. That’s AI as leverage: work that used to require hiring an external expert, done in-house, producing a lead you can invoice against. Compare that with my lost evenings perfecting a system nobody asked for. Same tools, same enthusiasm, completely different value.

The pattern behind the difference: AI pays off fastest when it replaces something you’d otherwise pay a consultant for or skip entirely, and when its output lands directly in one of your two pipelines, leads or people. It pays off slowest when you’re polishing internal machinery whose absence nobody had noticed.

And yet: you can’t sit this one out

Now the other side, because there is one, and it matters just as much.

We’re in a race right now. These tools are new, most of your competitors haven’t worked out how to use them, and the building businesses that learn to leverage them properly in the next couple of years will open a gap that’s very hard to close later. The builder who spends a few evenings a week getting AI running his admin, his follow-ups, his estimating support, is quietly getting ahead of the one watching Netflix.

My own version of this looks ridiculous and I stand by it: I’m on the couch at night, my wife’s reading her book, and I’m approving AI tasks by voice while we talk. She occasionally thinks I’m talking to her when I’m talking to Claude. A builder I coach does the same thing with his laptop on the couch, prompting between overs of the cricket. That’s not workaholism, it’s a few minutes of steering while something builds in the background, and it compounds.

So the answer isn’t less time with AI. It’s aim.

Three rules I’m trying to hold myself to

1. Start every session with the number. One sentence, before anything else: this session exists to move X. Sessions without an X get a time box.
2. Ship, then polish. The version that works and is ugly beats the version that’s elegant and three weeks away. My worst tinkering happens after something already works.
3. Do the fun stuff on purpose. If part of the work genuinely doesn’t feel like work to you, that’s not a trap, that’s information. Build the business so that IS your job, and hand the rest to systems and people. There’s no right way to build a business except doing more of what doesn’t feel like work.

AI is the most useful tool to hit this industry in my lifetime, and it’s also the most entertaining way ever invented for a builder to feel busy. The whole game is knowing, hour by hour, which one you’re holding.

If you want help aiming at the version that moves numbers, that’s precisely what my AI consulting sessions exist for.