Repair shop invoices, drafted by an agent
An auto repair shop handles about a dozen supplier invoices a week. A Claude Code agent on a Mac mini drives Chrome with no API. It opens supplier invoice mail, reads the PDFs, finds or creates the vendor, and saves a draft with amount and due date. In week six, a second flow began. It drafts customer invoices from finished work orders, with labour and parts lines. The owner reviews each morning and sends.
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 auto repair shop handles about a dozen supplier invoices a week. A Claude Code agent on a Mac mini drives Chrome with no API. It opens supplier invoice mail, reads the PDFs, finds or creates the vendor, and saves a draft with amount and due date. In week six, a second flow began. It drafts customer invoices from finished work orders, with labour and parts lines. The owner reviews each morning and sends.
In week one, the agent made two errors, both caught in the morning review. By 90 days, the combined error rate was under one in 20 entries. Morning review fell from 20 minutes to 10 minutes across both flows. The shop recovered nine to ten hours a month, at about $45 a month in running costs. Figures are the vendor's own case study, and its chart is labelled illustrative.
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.