How to win with AI

How to win with AI. If you're not technical, that might be an advantage.

Imagine you walk into a restaurant.

Declarative approach: "Table for two, away from the bar, please."

Imperative approach: "I've been here before. Grab two menus. Turn left. Walk ten steps. Turn right. Five more steps. Seat us there."

Both get you seated. But one trusts the system to figure out the how. The other micromanages every step.

When you're working with AI, be declarative. Focus on the goal, not the procedure.

Here's the trap I fall into with the technical areas I know best:

If you already know how to solve a problem, your instinct is to dictate 10 steps to the AI — when it could have reached the goal in 2.

I've watched non-technical people build mind-blowing things with AI — things they couldn't code 5% of themselves. That's not a bug. It's the point. They don't carry the baggage of knowing how it "should" be done. So they just describe what they want.

Real example from my tax prep this year:

I run an S-corp. My accountant needed docs to prepare my Form 1120-S and K-1.

My first instinct was imperative: "I need every transaction tagged correctly in QuickBooks. How can AI speed that up?"

Then I zoomed out.

What does my accountant actually need? About 15 data points.

So instead of automating a 50-step QuickBooks workflow, I used AI to write some automation that pulled from a couple of exports and produced exactly what my CPA needed.

Hours, not weeks. No subscription gymnastics.

The imperative mindset would have had me optimizing the old process. The declarative mindset helped me skip it entirely.

The principle:

When you bring a problem to AI, start with the outcome. Not "help me do steps 1 through 10 faster." But "here's what I need at the end — what's the shortest path?"

The declarative mindset doesn't just make you faster. It makes you dangerous.