This guide gives you a reusable starting pattern. It is designed to help you see the work more clearly; it is not a substitute for judgment, source checking, or responsibility for the result.
Prepare the inputs before you ask for output.
The model only sees what you give it. Spend a few minutes naming the reader, desired result, and uncertain information. This makes a first draft easier to assess and reduces the need for decorative rewriting later.
Give the task a useful brief.
Task: [what you need] Context: [relevant background and audience] Constraints: [what to include, avoid, or verify] Output: [format, length, and structure] Before drafting, tell me what information is missing or ambiguous. Then create a first draft and label any assumptions you made.
Replace every bracketed field with your real context. Read the output before reuse.
Four steps that keep the result usable.
- 1
State the task in one verb-led line.
- 2
Add only context that changes the answer.
- 3
Name constraints a reviewer would care about.
- 4
Ask for assumptions to be labeled so you can inspect them.
A prompt is a working brief
The useful part of prompting is not clever wording. It is the act of making a task reviewable. When you define reader, context, constraints, and useful output, you create a draft a person can assess with confidence.
Constraints protect the result
Constraints keep the model from doing work you never asked for. Examples include “do not invent citations,” “use only supplied notes,” or “leave facts uncertain if they are missing.” Put the important ones near the task.
A worked example: an internal memo
For a one-page memo on whether to change a pricing page, the brief might be: task, “draft the first section and a recommendation”; context, “readers are product and finance, decision by Friday”; constraints, “use only the attached usage data, no new numbers”. The model now has enough to draft something reviewable instead of a generic essay.
Write the brief before you prompt
The act of writing the brief is where most of the value appears. Naming the reader and the constraint forces the real questions to surface before any output exists. If you cannot write the brief in a few lines, the task itself probably needs clarifying first.
Run a human check.
- Can someone else understand the task from the prompt alone?
- Does the output format match how you will use it?
- Are important constraints explicit?
- Were assumptions identified and reviewed?
AI is strongest here when it makes missing information, structure, and options easier to see. The moment an output becomes a claim, commitment, or decision, bring a person back into the loop.