ThinkWithPrompt by SynotticThinkWithPrompt

Create a before, during, and after AI-use checklist for a team task

A short tick-box checklist covering tool choice, data handling, verification, disclosure, and sign-off whenever your team uses AI on one task.

Working with AIFew-ShotStandardWriteSummarizeFill-in Template

The prompt

Create a practical checklist for using AI on a task my team does regularly. The task: [e.g. "drafting job descriptions", "summarizing customer interviews", "preparing monthly sales commentary"] Who uses the checklist: [ROLE and AI experience level] Approved AI tools: [TOOLS] Data involved: [TYPES, flag anything sensitive] What has gone wrong before, or could: [e.g. "made-up numbers", "wrong tone with clients", "confidential details pasted in"] Build the checklist in three parts, each 5-8 tick-box items written as short actions: 1. Before - is AI appropriate here, approved tool, data cleaned or redacted, context and source material ready, clear prompt. 2. During - give sources rather than asking AI to recall facts, ask for assumptions to be stated, keep sensitive details out. 3. After - verify facts, numbers, names, and dates against sources; check tone and audience fit; check for bias or unfair statements; confirm a human owns the final version; disclose AI use if required; record what was done if needed. Then add: 4. Red lines - 3-5 things that mean stop and ask (e.g. "output will be used to decide about a person"). 5. A one-line version of the checklist for a sticky note or Teams or Slack channel header. Make every item specific to this task, not generic AI advice. Keep it short enough that people will actually use it.

Highlighted [brackets] are blanks to replace with your own details.

Fill in the blanks and use this promptMore Working with AI prompts

What to fill in

  • ROLE and AI experience level
  • TOOLS
  • TYPES, flag anything sensitive

How this prompt is built

This prompt follows the Few-Shot structure (Few-shot prompting). A few-shot prompt includes one or more worked examples of the input and the output you want, so the model can copy the pattern instead of guessing it.

Related prompts