← All frameworks ~ framework 03 of 05 · the selection tool

Before you build, you score.

Most operators waste their first year of AI on a task that was never going to work. They pick the most visible task — the email, the proposal, the thing their team complains about most. It sounds like the right move. It almost never is.

Here’s the thing. The right dish to dispatch first isn’t the one that feels annoying. It’s the one that scores highest on four specific dimensions. Four numbers. One out of twenty. If you want to stop guessing which system to build next, you score every candidate on the same four questions.

That’s the Prep List.

Brass measuring cups arranged in ascending size on a weathered wooden cutting board — the ritual of measurement before you cook.

measure twice ~ cook once

~ the selection tool

The Prep List, in one scorecard.

Score every candidate dish on four dimensions. Total out of twenty. The verdict tells you whether to write the recipe now, queue it, or leave it with the Chef.

01

Repeatability

Does this dish come up on the menu daily — in near-identical form? Or is every plate a custom build?

rare / custom
daily / identical
02

Volume

How much of your week does cooking this eat? Hours a week is different from hours a month.

minutes
multiple hours/day
03

Definability

Can you clearly articulate what “plated well” looks like? Could you describe it to a new hire on day one?

fuzzy / judgment-only
written standard
04

Reversibility

If the station burns the plate, how easy is it to fix? Embarrassing email vs. losing a customer.

irreversible
trivial to fix
Total ____ / 20
17–20
Write this recipe first
High-leverage, low-risk. This dish leaves the Chef now.
12–16
Good candidate
Queue it behind the top-ranked dishes. Write the recipe next.
4–11
Keep with the Chef
Not ready for the line — either too fuzzy or too high-stakes. You still cook it.
~ four dimensions, one out of twenty

The four dimensions, expanded.

Repeatability asks whether the dish comes up on the menu daily in near-identical form. One-off custom work is hard to systemize. Daily repeatable work is an open door. A 5 here means you could describe every instance of this task in the same three bullets.

Volume asks how much of your week it eats. Because leverage comes from the hours you get back. A task that takes three minutes a week isn’t worth a recipe, no matter how repeatable it is. A task that eats four hours a day is a gold mine even if you score it a 3 on everything else.

Definability asks whether you can clearly articulate what “plated well” looks like. Could you describe it to a new hire on day one? If the answer is “I’ll know it when I see it,” the station can’t cook it. The recipe doesn’t exist yet.

Reversibility asks what happens if the station burns the plate. An embarrassing email you can fix is not the same as a price quote that blows a contract. High-stakes, irreversible work stays with the Chef until the recipe is rock solid.

The verdict brackets.

17–20. Write this recipe first. High leverage, low risk. The dish leaves the Chef now.

12–16. Good candidate. Queue it behind the top-ranked dishes.

4–11. Keep with the Chef. Either too fuzzy or too high-stakes for the line. You still cook it — for now.

The right dish to dispatch first isn't the one that feels annoying. It's the one that scores highest.

How it connects.

The Prep List picks the dish. Everything downstream runs on that pick. The Prep List feeds the Four D’s — a 17+ dish is one you can actually move from Doing to Dispatching this week. It feeds the Professional Recipe — a high-scoring dish is one a recipe can actually contain.

Most operators waste their first year of AI on a dish that scored an 8. They built the recipe. It didn’t hold. They concluded “AI doesn’t work in my business.” The tool wasn’t the problem — the dish selection was.

~ the monday move
Inventory five repeated tasks this week. Score each on the four dimensions. The highest score is your next recipe.