200 transactions reconciled, 11 wrong
An owner of a few small businesses stopped pasting bank files into a chat. He had Claude Code build a bookkeeping pipeline he can run again each month. The pipeline pulls the month's transactions from his accounting platform, matches each bank row to a receipt or invoice by amount and date, and proposes a category for every row with a one-line reason. It flags any row it is unsure about instead of guessing in silence.
Start with one job only. Fill in the three lines. Keep the bot on that job.
The one job this bot does: ______________________________
What it may do on its own: ______________________________
What it must ask me first: ______________________________
Two lines from this case, side by side. No new figures.
An owner of a few small businesses stopped pasting bank files into a chat. He had Claude Code build a bookkeeping pipeline he can run again each month. The pipeline pulls the month's transactions from his accounting platform, matches each bank row to a receipt or invoice by amount and date, and proposes a category for every row with a one-line reason. It flags any row it is unsure about instead of guessing in silence.
For the month he reports, the pipeline reconciled 200 transactions and miscategorized 11, a 94.5% hit rate. His review fell from re-checking 200 rows to checking the 18 flagged rows plus a sample. Seven of the 11 errors sat in the flagged pile. The pipeline never added wrong, lost a row, or counted a row twice.
The after line is the source's claim, not our measurement.
smalloffice.space · Get a case like this in your inbox each morning. Subscribe free
Read this case on your phone now, and keep it for later. Type your email. We send this case to that address, and you get the next cases each morning too.