When iteration is free, deciding is the moat

Jeff Gothelf wrote a sharp piece this week called Three Habits That Beat AI Tool Fluency. It's framed for product designers and product managers, but the core point applies to anyone running a business right now.

The short version: when AI makes iteration free, polish stops being a stop signal. You can keep making things "better" forever. The teams that win aren't the ones with the best prompts. They're the ones who know when to stop.

Here's the operator translation.

What's actually changed

Six months ago, getting a polished version of anything took real time. A landing page concept. A draft email sequence. A marketing brief. A dashboard mockup. Hours, sometimes days of work to produce one option.

Now you can produce three polished versions in an hour. Anyone on your team can. Your client can. The teenager you hired to help with social can.

That sounds like a win. But there's a hidden problem nobody talks about: when iteration is free, the natural stop signal disappears. You used to stop because the work was hard and you were tired. Now you stop because someone says enough, or because the meeting ended, or because you got bored. None of those are the same as stopping because the work is actually done.

Gothelf's insight is that this changes what's valuable. Tool fluency becomes table stakes. Decision quality becomes the moat.

The three habits

Gothelf names three habits that produce decision quality. Worth reading his full piece, but here's how they translate to running a business:

1. Name the outcome before you write the prompt.

"A better landing page" isn't an outcome. "More qualified leads from the homepage" might be. "Visitors book a discovery call within 30 seconds of landing" definitely is.

If you can't articulate the outcome specifically, what you have is a wishlist. AI will generate beautiful versions of your wishlist forever and you'll never know if you're done.

2. Decide the stop criteria up front.

Before you start iterating on anything, write down what would make you stop. We stop when the page tests with five customers and three book a call. We stop when revenue from this campaign exceeds our cost. We stop Friday at noon, regardless of where we are.

Any explicit stop criterion beats the implicit one. Without one, you're running a polish factory.

3. Design the test that would prove you wrong.

Every iteration is an implicit hypothesis. Your job isn't to be impressed by the iteration. Your job is to figure out the cheapest test that would tell you the direction is wrong, and run it before producing another version.

Why this matters for operators

If you're running a small or growing business, here's the trap waiting for you in 2026:

You hire a marketing person. Or a designer. Or a contractor for some piece of work. They use AI tools. They produce three polished versions of everything you ask for. They show you all three. They ask which one you like.

You like all of them. Or you can't tell which is better. Or you go back and forth picking different ones on different days.

This is the polish factory. You're paying someone to iterate beautifully on something nobody decided what success looks like for. The output looks impressive. The outcome doesn't move.

The fix isn't more iteration. It's the three things on a piece of paper before anyone touches a tool: what's the outcome, what's the stop signal, what's the test.

The practical version for next week

Pick one thing your team is currently iterating on with AI. Before the next session, write three things on a sheet of paper:

1. The outcome this iteration is supposed to serve, in your specific terms. If this works, what will customers be doing differently?
2. The specific criteria, ideally a number, that would tell you to stop iterating.
3. The test that would prove the current direction wrong or right.

Then run the next session with those three things on the table. The conversation goes from "which one is prettier" to "which one moves the metric." You'll be surprised how fast you reach a decision.

Tool fluency is the price of admission. Deciding is the moat.

Read Jeff's original piece →