Use four pieces of information

Most disappointing prompts are missing context. A short request can still be specific if it names the situation, the task, the reader, and the standard.

  • Situation: What is happening, and what does the tool need to know?
  • Task: What should it draft, compare, summarize, or organize?
  • Reader: Who will receive or use the result?
  • Standard: What tone, length, format, and limits matter?

Give source material when accuracy matters

If you want a summary, paste the text or attach the document when your tool and workplace policy allow it. If you want an email, provide the points that must be included. Asking a model to rely on your source material is safer than asking it to fill gaps from memory.

Add a simple instruction: “Use only the information below. If something is missing, list the question instead of guessing.”

Correct the draft with ordinary language

You do not need to restart when the first answer misses. Say what is wrong: “This is too formal,” “The second paragraph invents a reason,” or “Put the decision before the background.” Specific feedback usually works better than adding more prompt jargon.

Practise on real work

Turn rough notes into a useful brief

Choose notes that contain no confidential information.

Prompt to adapt

Situation: [two sentences]. Task: turn my notes into [output]. Reader: [person or group]. Standard: [tone, length, format]. Use only my notes. Put missing information under a heading called Questions before finalizing.

Judgment checks

  • Did the tool separate facts from missing information?
  • Is the requested action easy to find?
  • Can you remove any sentence without losing meaning?

Keep these points

  • Context beats clever prompt formulas.
  • Provide the facts instead of inviting guesses.
  • Correct weak output in plain language.