As of 13 August 2026, AI can find missing transactions in your bookkeeping.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, power-user skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Open the business bank account and export the relevant transaction period as CSV or download clear statements covering the same dates.
   2. Open the bookkeeping software and export the transaction ledger, including dates, amounts, descriptions, references, account or category and VAT fields where available.
   3. Remove unrelated personal accounts and redact unnecessary personal information, then check that both exports use the same currency and cover the same start and end dates.
   4. Paste the prompt into a chatbot and attach or paste both exports, labelling one as BANK DATA and the other as LEDGER DATA.
   5. Ask the chatbot to rerun the comparison if it reports missing columns, unreadable rows or a mismatch in opening and closing balances.
   6. Compare every high-confidence candidate with the original bank statement and supporting invoice or receipt, then correct only confirmed omissions or duplicates in the bookkeeping software and send unclear or VAT-sensitive items to your accountant.

   Prompt

   I need to find transactions that may be missing from my business bookkeeping for [business name] between [start date] and [end date]. I will provide:
   1. A bank transaction export or statement.
   2. A bookkeeping ledger export.
   
   Compare the two sources by date, amount, currency, transaction type, reference and counterparty. Account for differences in date format, bank processing dates, duplicated imports, transfers between my own accounts, refunds, card payments, cash withdrawals, pending items and opening or closing balance adjustments.
   
   Do not invent transactions, categories, VAT treatment or explanations. Do not post or amend anything. Produce a table with these columns: bank date, value date if available, amount, description, bank reference, whether a matching ledger entry exists, possible ledger match, reason it was flagged, confidence as high or low, and what I should check next. Separate the results into likely missing transactions, possible duplicates, timing differences and items that need a human decision.
   
   Show the reconciliation totals for the bank data and ledger data, explain any difference, and identify missing rows or incomplete data that could make the comparison unreliable. Treat a transaction as a candidate only, not as proof that it is missing. Finish with a short list of the source rows I must inspect manually before making any bookkeeping change.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.