Find the recurring issues in tickets
Same complaint again. Different ticket number, same root-cause meeting.
Tested Claude Opus 5.5, medium effort · 15 Nov 2026 · on the practice files
Get week's tickets
Use unique ticket ref, category and anonymised note. Missing category stays blank.
Privacy Remove customer phone numbers, addresses, order payment details and personal identifiers. Use ticket or order refs instead.
Whenever I send ticket export for [Period], apply [Counting rule]. A. Count categories and percentages, include Unknown. If rounded percentages do not add to exactly 100%, say so (rounding). B. List examples by ticket ref; separate symptoms from confirmed causes. C. Three suggested follow-up checks, not decisions, each with a suggested owner role marked [confirm]. Treat a cause as confirmed only when my rule's test is met (for example completed job-card evidence, from the file or a record I supply); without such a rule, call it a hypothesis until the owner confirms the root cause. Do not name a person as at fault. Use only supplied facts and policy. No admission of legal liability, refund or replacement promise unless the supplied policy permits it and an authorised person confirms. No invented dates, prices or outcomes. A person checks before sending or deciding.
Set it up once. Save the approved rules and output format, not customer details. Pick your tool at the top to see how.
What is that?
A Project is a folder in ChatGPT that keeps your instructions and files for every chat inside it. It's on the free plan too. Checked 14 Nov 2026.
What is that?
A Project in Claude is a workspace with its own instructions and files that every chat inside it uses. Checked 14 Nov 2026.
What is that?
A Gem is a custom Gemini that loads your instructions before every chat. Free on every plan; you create it on a computer. Checked 14 Nov 2026.
What is that?
Signed in with a work account, Microsoft's Copilot Chat applies enterprise data protection: your prompts aren't used to train its models. Checked 14 Nov 2026.
After that, each time: send the next masked notes or export and current rules; check the result.
First time? Try it in a normal chat: paste and send.
Which model? Use your tool's strongest model with thinking turned on.
In your AI tool: attach the files with the + button, paste, send.
- Delay3 of 6. 50% of sample.
- Wrong item2 of 6. Reported symptom.
- T6Missing category. Ask agent.
Then send this: Recount categories and percentages; confirm every example from file.
Guesses cause
Reply: "Symptoms and checks only."
Hides unknown
Reply: "Count Unknown."
Counts twice
Reply: "One category per ticket."
Sir ne bola urgent. Ab bolo: ho gaya.
Did it work?
If your whole team does this, we'll set it up for all of them in one session.