LANDMARK · 5 MIN · USE
A useful request
After this landmark, you can turn a vague ask into a request the model can actually act on, by stating what it otherwise has to guess.
A model can’t read your mind, and a one-line prompt leaves it filling the gaps with the statistically average guess — which is rarely what you wanted. The fix isn’t longer prompts; it’s naming the five things a competent stranger would ask before starting: goal, context, audience, boundary, and format. Goal is the outcome you want; context is the background only you have; audience is who reads it; boundary is what to avoid or how far to go; format is the shape of the answer. Say those, and the model spends its effort on your problem instead of inventing a generic one.
This is the canonical concept. It stays the same across learner lenses so personalization never changes the underlying facts.
◌ EVERYDAY LENS
What this looks like for you
Before you hit send, reread your prompt as if a helpful stranger got it. Would they know who it’s for and what “done” looks like? If not, add that one missing line.
MAKE A DECISION
You ask an AI to “write a summary of this report” and get back something generic and too long. What’s the best next move?
CARRY THISTake one prompt you’re about to send and add the single line it’s missing — usually audience or format.