Ledger entries that need a closer look
'Miscellaneous expenses': ₹14 lakh
Tested Claude Opus 5.5, medium effort · 15 Nov 2026 · on the practice files
Export relevant ledger lines
Use voucher ref, date, ledger head, amount and supplied approved expected head if available. Include evidence status.
Confirm review rule
Use your own threshold and approved mapping. A flag is not proof of wrong posting.
Privacy Use voucher refs and business amounts; remove personal identifiers, bank details and confidential narration. Delete bank account numbers; replace customer UPI IDs and phone numbers.
I'm attaching ledger lines for [Review period]. Threshold ₹[Review amount above]. Checks: [Checks]. A. Per row flag repeated voucher ref (all occurrences), actual vs approved expected head mismatch, evidence missing and amount above threshold. Show reasons separately; don't infer wrong head if expected unknown. B. Sum all supplied amounts and flagged amounts once per row; list exact source refs for review. Keep duplicates in totals until verified. C. Questions for accounts and proposed review sequence. No entry correction or tax advice. Use only my data. GST rates, HSN and TDS rates only from my supplied figures; never guess tax treatment. This is a first pass; a person checks and posts any final entries in Tally/Busy.
Which model? Use your tool's strongest model with thinking turned on.
In your AI tool: attach the files with the + button, paste, send.
Keep the actual records in your own Excel workbook; share only headings and rules.
Excel A:F voucher, date, actual head, amount, approved head, evidence. Threshold ₹[Review amount above], checks [Checks]. Give formulas for all duplicate refs, actual vs supplied approved head, missing evidence, above threshold and combined review flag; sum every source amount and flagged rows once. Tell me where to paste formulas or how to set up the local steps; use fictional examples only, no actual records.
Run locally, verify flagged rows against your source and have the responsible person approve any action.
- V2Repeated ref. Both ₹10,000 occurrences flagged.
- V3Head mismatch. ₹5,000 booked to repairs vs approved stationery.
- V4₹25,000. Above threshold; evidence missing too.
Then send this: Recount rows, duplicates and flags, sum all and flagged amounts, verify expected head only from source.
It deletes duplicate
Reply: "Keep in total; verify first."
It invents correct head
Reply: "Only supplied approved head."
It treats flags as fraud
Reply: "Possible issues, not allegations."
Sir ne bola urgent. Ab bolo: ho gaya.
Did it work?
Doing this every week? Save it once.
Save the prompt in your AI tool, so next time you only drop the files in. 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.
If your whole team does this, we'll set it up for all of them in one session.