Critical Thinker's Playbook

for Day-to-Day AI Use in Business

Clear, practical guidance for professional AI tools

🔷 BEFORE PROMPTING ▼

1. Framing the Need

Be clear about what you want and why

▼

Questions to ask yourself:

  • What decision or task am I trying to complete?
  • Who will read or use the result?
  • What practical limits matter (time, resources, deadlines)?
  • What form should the answer take?
Example:
"I need help choosing which customer feedback themes should shape our next release. The result should help me explain the decision to the engineering team."

2. Crafting the Prompt

Turn your need into a clear request with context and limits

▼

How to do it well:

  • State the situation
  • State the goal
  • State the limits
  • State the form you want the answer in
  • Ask for assumptions to be included
Example prompt:
"Given our goal to reduce customer frustration during the setup process, list the top three improvements based on these customer comments. Keep suggestions achievable within six weeks. Provide a short explanation for each."

3. Planning Reproducibility

Decide ahead how you will test response stability

▼

Simple ways to test:

  • Ask again using slightly different wording
  • Ask for the same content in a different format
  • Compare results from another AI system or source
Example:
"I will ask for the list once as bullet points and once as a short paragraph. The priorities should be similar."
🟩 AFTER PROMPTING ▼

4. Assess Completeness & Relevance

Make sure the answer covers everything you asked for

▼

Checks:

  • Does the response address the full question?
  • Did it include the requested format?
  • Is anything missing or off topic?
Example:
"If I asked for three improvements but only got two, I need a revision."

5. Assess Accuracy & Credibility

Check whether the information is believable and supported

▼

How to check:

  • Compare claims to reliable internal data or feedback
  • Look for statements that sound certain but lack support
  • Ask the tool to list assumptions and uncertainties
Example:
"If it claims 'users always struggle with payment registration,' I check our support tickets or surveys to confirm if that's actually true."

6. Assess Currency & Context

Ensure the answer reflects the current situation

▼

How to check:

  • Ask what parts might be out of date
  • Compare with recent changes or market conditions
  • Ensure suggestions match current product and customers
Example:
"If it recommends improving a feature we removed last month, it's not current and needs correction."

7. Understand Implications

Look at consequences of using the response in real work

▼

Questions to ask:

  • What would happen if I followed this suggestion?
  • Who would be affected? What risks or trade-offs exist?
  • What might go wrong if the response is incorrect?
Example:
"Would prioritizing this delay a promised integration for a key partner?"

8. Treat Output as a Draft, Not Truth

Remember that AI gives useful starting points, not final answers

▼
✓ Edit for clarity and correctness
✓ Add your own judgment
✓ Confirm important parts with real data
✓ Make the final decision yourself