The AI bookkeeper files bank lines with built-in keyword rules before any learning happens. To find where they are thin, we write descriptors for a kind of business not yet covered and count what the rules do with them.
This round covered florists, house painters, handymen, freelance translators and process servers or court reporters. Of 42 descriptors, the rules filed 13 right, 2 wrong and 27 not at all. We dropped or corrected three expectations of our own: a Wise transfer is a transfer, not income; a cooler repair has no account in the default chart; and payroll is wages, not an operating cost.
The 40 that remained now file correctly: floral wholesalers and paint makers to supplies, translation tools to software, trade associations to dues, and wire-service and translation-marketplace payouts to sales. That is tuned on the same list, so it counts coverage, not accuracy. The descriptors are made up from public company names; no real bank export was used. Labelled real books are still what the project needs most.