Library › Customer Support & Success AI › Customer success managers
Build a promise ledger in CSV with source type and status
- For Customer success managers
- You practice Asking for a format you can use
- Time about 5 minutes
- Level Advanced
About this exercise
Return one code block with CSV using the header promise,source,status, one row per promise in the order they appear. source is exactly "contract", "email" or "verbal". status is "open" unless a later item says it was fulfilled or withdrawn (then "done" or "withdrawn"). Do not say which promises are binding. Write no more than 6 words outside the block.
Common mistakes when you ask AI for this
- Treating everything as binding. The AI follows Marta's note. Try: Say: "Do not say which promises are binding; just record source and status."
- Missing the withdrawn promise. The AI lists the go-live as open. Try: Say: "Use the later update: withdrawn."
- Wrong source label. The AI uses written instead of email. Try: Say: "source is exactly contract, email or verbal."
- Skipping the code block. The AI writes prose around the data instead of one block. Try: Say "return exactly one code block".
- Letting the block sprawl. The AI pretty-prints and adds extra lines. Try: Say "keep the block under N lines".
- Talking around the block. The AI adds an intro and notes outside the block. Try: Say "no words outside the block except ...".
- Leaving a requirement unstated. The AI follows its own defaults unless the prompt spells this out. Try: State it plainly in your prompt: the code block has the header promise,source,status and exactly 8 data rows (the two order form items, the Shopify connector, the whitepaper, the references, the discount, and the Dec 1 go-live) with no more than 10 lines in total.
What a strong prompt includes
- Gives the header
- Source values
- Status rule
- No binding talk
- Asks for one code block
- Keeps the block short
- Limits text outside the block
Related exercises
- Acknowledge a cancellation request without a save attempt (Customer Support & Success AI · Customer success managers)
- Acknowledge an accessibility bug without a promised date (Customer Support & Success AI · Customer support teams)
- Acknowledge an executive escalation within the hour (Customer Support & Success AI · Executives)
- Adapt one ad image to 3 sizes with Midjourney (Image & Video AI · Marketing and content)
- Age a returns queue in business days and flag the ones over the limit (Customer Support & Success AI · Customer support teams)