In this lesson you will
Write a brief that gives an AI assistant a clear goal, audience, approved facts, format, and limits, then judge the draft against that brief instead of against how confident it sounds.
The idea in plain language
When an AI draft sounds generic, adding a dramatic instruction rarely helps. A clear brief does. OpenAI's own guidance is to be clear and specific, give enough context, and refine the prompt after reviewing the response.
A useful brief has five parts:
- Goal: what the reader should understand or do.
- Audience: who they are and what they already know.
- Approved facts: only details you have confirmed, plus what is still undecided.
- Format and tone: length, structure, and voice.
- Boundaries: what it must not invent or promise.
Worked example
IllustrationSplitting shared bills with roommates (fictional)
A vague prompt like “write a message about splitting bills” invites the assistant to make up amounts and rules. The brief below fixes that.
Write a short message to my two roommates proposing how we split shared monthly bills. Audience: two roommates who already split rent equally but not utilities. Goal: propose a simple way to split internet and electricity, and ask them to agree or suggest changes. Approved facts: internet and electricity are the only shared bills; the amounts change each month; nothing is decided yet. Format: one short paragraph, then one question. Tone: friendly and plain. No pressure. Boundaries: do not invent dollar amounts, due dates, payment apps, or rules we have not agreed. Keep it under 90 words. After the draft, list the facts I must fill in before sending it.
When you review the draft, look for anything the assistant added on its own, such as a specific amount or a deadline. Cross it out or replace it with a fact you can confirm.
Try it: revise once
- Run the practice prompt in an AI tool you already use.
- Point to one sentence that misses the brief and ask for a focused change, for example: “Make the second sentence shorter and remove the deadline.”
- Try the same brief with a different tone. Keep the version that fits the reader, not the most enthusiastic one.
Check the result yourself
Limits of this lesson
- A fluent draft is still a draft. It does not prove any detail is correct.
- AI tools change often. Features and settings described in sources can differ in the tool you use.
- Use fictional or low-stakes examples first, and remove private details before pasting anything real. The privacy lesson covers how.
Keep the human review
Examples are practice exercises, not tested AI outputs or a promise of results. During practice, do not give an assistant permission to send, publish, buy, or change anything in an account.
Sources
Facts on this page were checked against these sources on October 7, 2026. Figures marked Illustration are arithmetic examples, not forecasts.
- OpenAI Help Center, Prompt engineering best practices for ChatGPT (checked October 7, 2026) (opens in a new tab)
Be clear and specific, give context, refine iteratively.
Go deeper
Related articles from the MentorSurge archive. Articles reflect the date they were published; check time-sensitive figures.
- AI Literacy Is the New Financial Literacy: A 12-Month PlanMentorSurge article · published June 1, 2026
About this lesson: original MentorSurge teaching material for general education. It is not individualized financial, investment, tax, or legal advice. Spotted an error? Tell us.
Turn Rough Notes into a Clear Plan
How to use AI to structure notes into an action plan while keeping unknowns visible.
How to Read the MentorSurge Market Dashboard
How to interpret ETF quotes, percentage changes, quote times, and a composite market score.