Vol.057 – Week 5 of Going AI Native: Connecting the Skills Was Easy. What It Revealed Wasn’t.
Date: 2026-07-10 | Newsletter
Key Summary
Week 5 of DCC Corporate Services’ public 12-week AI Native project was, Zenta says, the most useful yet — precisely because nothing broke. The four individual accounting skills connected into one sequence, and two of his accountants produced real, uploadable output: one turned fifty-plus supplier invoices into a single master file built batch by batch across the week; another turned a full month of expense claims — foreign-currency conversions and entertainment flags and all — into a clean, coded file. But connecting the work exposed what no single skill could. First, two strong performers structure work differently — many small per-claim files versus one ever-growing master ledger carrying its own confidence-level columns — friction between two good habits, not between a strong worker and a weak one. Second, the exit: the legacy accounting software takes the better part of ten clicks to load an upload, so all the new upstream speed may just pile up in front of the bottleneck (Goldratt’s “The Goal”). The real output of Week 5 isn’t a finished pipeline — it’s a decision Zenta now has to make with his eyes open: standardize everyone onto one method, or let each person peak her own way. His own goal argues with itself — “10x capacity” leans one way, “consistent quality across every client and five languages” leans the other. A clear goal, he argues, is what turns a friction into a decision instead of a stall. Do it simple. When the work finally connects and starts to rub, go back to your goal before you go back to the rulebook.