Low-hanging fruit first

Ask ten consultants how to apply AI to a business and you’ll get ten visions. Every AI pitch I see starts with a vision. Mine start with a list of what went wrong last month.

I’ve been automating business processes for nearly thirty years — at a bank where the stakes were board reports, and in my own companies where the stakes were my own sleep. In all that time, one approach has never failed me, and I’ve watched the opposite approach fail over and over. Here it is.

How to apply AI to a business: go after the low-hanging fruit

When you bring AI into a business, you don’t start with a grand vision. You start with the pain.

First, the things that keep the owner up at night. The near-misses. The thing that went wrong once and could have really hurt you, so now you sleep with one eye open. Here’s the secret nobody tells you: these are the easiest problems to solve with AI. They feel big because they’re scary, but they’re almost always specific and repeatable — and specific, repeatable checking is exactly what machines do better than people. I wrote a whole piece on this: start with the 3am list.

Second, the operational friction. The stuff that just takes longer than it should. Figuring out schedules. Matching invoices to payments. Flagging booking conflicts. Onboarding a new client. None of it is dramatic. All of it is a quiet tax on every single week — and AI knocks these out fast. That’s part three.

Then — and only then — the bigger ideas. Once the foundation is in and the owner trusts the systems, the ambitious projects become straightforward to layer on top. Skip the foundation and the ambitious project becomes a crater.

The only ratio that matters

Every project I take on gets judged by one measure: the most improvement to the owner’s life and the business’s processes for the least effort. Relief-to-effort. Not impressiveness. Not novelty. Not what demos well on stage.

A watchdog that catches double-bookings before a guest is standing in an occupied condo takes days to build and removes a fear the owner has carried for years. A ground-up “AI transformation” takes a year, costs six figures, and usually under-delivers — Gartner predicts at least 30% of generative-AI projects get abandoned right after proof of concept, citing “unclear business value” among the leading causes. Unclear business value is analyst-speak for: nobody started with the pain. I know which one I’d rather ship, and I know which one the owner actually needed.

Why consultants get this backwards

Because most of them have never owned anything. If you’ve never had a business wake you up at 3am, you don’t know what the pain points feel like — so you reach for big shiny ideas instead, and next thing you know there’s an AI running amok in a business the builder never understood. That’s the entire difference between the projects that fail and the projects that pay — not the model, not the budget, but whether the person building it has actually run a business. The technology is available to everyone. The thirty years of knowing where businesses actually bleed is not. That’s part four, and it’s the contrarian one.

The sequence

Prove the concept. Make the day-to-day easier. Earn trust. Then go up from there. It’s not a limitation — it’s the fastest route to the big stuff, because every small win funds the next one. Part five closes the series with how that compounding actually plays out.

None of this is complicated. It just requires having felt the problems you’re solving. Everything else on this site — the client work, the services — is this one idea, applied.