From messy meeting notes to a verified decision record

An advanced free sample course where you build, review and stress-test an AI assistant in about 15–20 minutes.

You receive fictional meeting notes and an AI summary that looks professional. The problem is that the summary mixes facts, proposals and invented details. Your mission is to make the workflow safe enough to use after the next meeting.

How it works

1. 1Review an AI response that appears credible.
2. 2Identify what the meeting notes actually support.
3. 3Choose rules for a better assistant.
4. 4Stress-test it before taking the instruction with you.

What you will learn

• Spot when a convincing AI response contains assumptions or fabricated details.
• Distinguish decisions, action items, proposals and open questions.
• Create rules that make a reusable assistant safer and more consistent.
• Stress-test the assistant against conflicting information, hidden instructions and sensitive data.
• Take away a ready-to-use project instruction for ChatGPT Projects or Claude Projects.

You do not need to know any technical terminology. Every task tells you exactly what to do and why.

Use only the fictional material in this exercise. Do not paste personal data, customer information or other sensitive information.