Resource · Lesson 1

Context checklist.

A two-part reference

The 4-part context method for writing a better prompt, and the 6-question verification for checking the response. Print it. Reuse it.

Part 1 · Before you send the prompt

The 4-part context method

Use these four layers in every prompt that needs to produce something usable. Each layer is a small thing the AI needs in order to give you a good answer — skip one and it has to guess.

  1. 01

    Role

    Tell the AI who it should act as.

    The role gives it a lens. A project manager, an HR partner, and an executive assistant would all summarize the same meeting notes differently.

    You are an experienced project manager.
  2. 02

    Task + Format

    Say exactly what to do and what shape the output should take.

    Without this, the AI decides what "done" looks like. "Make this better" is not a task. "Turn these notes into a numbered list with owner and deadline per item" is.

    Turn the meeting notes below into a list of action
    items. For each: owner, deadline if mentioned, and
    a one-sentence description. Put urgent items first.
  3. 03

    Example

    Show one example of the output you want.

    An example does most of the heavy lifting for format and tone. It is faster to write than instructions and harder to misinterpret.

    Example style:
    - Sarah — draft launch checklist — Wed — needed for kickoff
    - Marco — confirm vendor pricing — Fri — blocking budget review
  4. 04

    Details

    Paste the source material the AI needs to work with.

    The AI cannot read your screen, your inbox, or the doc on your desk. If you want it to act on something, the content has to be in the prompt.

    Meeting notes:
    [paste your notes here]

Part 2 · Before you use the output

Verification checklist

Run through these six questions every time you ask an AI to do work for you — before you send the prompt and again before you act on the reply.

  1. 01

    Does it answer the question I actually asked?

    Not a related question. Not a fancier version. The one I asked.

  2. 02

    Is every fact something I can verify in my source material?

    If you did not write it and the AI did not cite it, you cannot trust it.

  3. 03

    Are owners, dates, and numbers real and complete?

    Not "TBD" unless you accept "TBD" in the final answer.

  4. 04

    Is the format what I asked for?

    Checklist when I asked for a checklist. Email when I asked for an email.

  5. 05

    Are urgent items at the top, not the bottom?

    The AI does not know your deadlines unless you said so.

  6. 06

    Did the AI invent, infer, or assume anything I did not provide?

    If yes, fix it before sending or mark it clearly as a guess.

Want to see this in action? Lesson 1 walks you through both parts step by step.

Start Lesson 1 →