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 freelance editors, podcast producers, virtual assistants, farriers and bounce-house renters. Of 34 descriptors, the rules filed 14 right, 3 wrong and 17 not at all. The three "wrong" were our own expectations: microphones and cameras are equipment, and an equipment-rental house files under repairs and rentals. The real gaps were editing and virtual-assistant marketplace payouts (Reedsy, Fancy Hands, Time etc), style and grammar tools, trade dues (editors, podcasters, farriers, virtual assistants, inflatable operators), farrier and inflatable supply houses, an e-signature tool and an event-lead fee.
After the rules, all 34 descriptors file correctly. 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.