Sample lesson
Writing prompts that work
About 8 minutes to read. This is the lesson exactly as a member sees it.
Most people blame the AI when the output is poor. The problem is almost
always the instruction.
An AI has no memory of you before a conversation starts and no access to
anything you have not given it. Everything it knows in that session comes
from what you write. The quality of what you put in determines the quality
of what you get out.
Three reasons prompts fail. They are too vague, so the AI guesses and guesses
wrong. They carry no context, so the AI does not know your company, industry,
audience or constraints. And the first draft gets accepted, when the
professionals who get the most value treat the first answer as a starting
point and refine it through conversation.
CLEAR fixes all three. You do not need all five elements every time, but when
you are not getting what you need, working through each one almost always
finds the problem.
**C is Context.** The situation. Your role, the industry, the specific
problem, the constraints and what has already been tried.
**L is Length.** How long and in what format. Be specific: exactly three
sentences, 200 words, a table with three columns. Without this the AI
rambles.
**E is Examples.** Show it what good looks like and it will pattern-match to
it. Match the tone of this email, then paste one.
**A is Audience.** Who reads the output. A chief executive cares about
return, a risk officer about compliance, a line manager about team capacity.
The same facts need different framing.
**R is Role.** Assign a specific expert identity. You are a senior commercial
lawyer in Zambia with fifteen years in mining contracts. This is calibration,
not manipulation. It narrows which patterns and vocabulary the model draws
on.