The reason AI hands you a rough draft is almost never the model. It is the brief. Given a thin instruction, the model fills the gaps with generic, average assumptions. A new staffer would stop and ask a few questions first; the model does not ask, it guesses, and it guesses average. This prompt makes you brief the model the way you already brief a person.
Use it on any deliverable you run regularly and keep having to fix by hand: competitive research, a go-to-market plan, a report, a client summary. Best for the operator tired of spending two hours cleaning up what should have come back finished.
The one-line request you would normally type, plus a sentence on who the output is for and what you will do with it.
You are my delegation editor. I am about to hand a task to an AI the way I would hand it to a new hire. Here is my rough request: [PASTE].
Before I send it, interview me to build a real brief. Ask me, one at a time:
Wait for my answers. Then assemble a clean, four-part brief I can paste into any model, and flag the one part I left thinnest.
Answer the four questions honestly, paste the assembled brief into your model, and compare the output to what your one-liner used to produce. Save the brief; most of it is reusable next time.
A four-part brief (context, deliverable, standard, example) tight enough that a stranger could run the task and get close to what you wanted.
Skipping the standard, which is where average output hides. Giving no example, so the model guesses the level. Treating the brief as one-time instead of a reusable asset.