This week’s pieces kept finding work that a rule or a decision had created and that nobody was assigned to look at, from the learning still left inside a program granted the fast lane to a birthday card waiting on a store manager’s list.
On October 14 the Federal Reserve released its October Beige Book, a summary of what business contacts told the twelve regional Reserve Banks on or before October 5. Most Districts described little or no net change in employment. Five reported that some firms had reduced staff through attrition, delayed hiring or expected to create fewer jobs after raising productivity by investing in AI. In the Dallas District, a staffing firm said it hired new recruiters to help manage its AI expansion, because AI agents had conducted more than 180 hours of interviews in two days and people still had to review them and make the decisions.
Among the AI anecdotes in the twelve District reports, only that firm describes the work the extra output created. It matches this week’s piece on a University of Auckland preprint, where engineers using AI assistants said their effort tilted toward checking, work the authors say standard task categories miss.
My read, which goes past anything in the Beige Book, is that a staffing plan built on AI productivity repeats the error in this week’s piece on the Pentagon’s fast pathway. The decision is made on the promised speed, and the learning about what the tool actually demands happens afterward, inside a smaller team. Attrition hides that well. Nobody announces a role that goes unfilled, so the review hours turn up in the weeks of whoever stayed, often people still paid and promoted for what they produce themselves, the same gap the factory study found in team leaders paid by the piece.
The Beige Book collects anecdotes and cannot say how common either pattern is. It does show firms making headcount calls on AI productivity this fall, and my guess is that few of them have measured the checking yet.
If you have not heard Episode 5 of High Road Conversations, Otis Florence went into business for himself straight out of high school and has spent about thirty years helping people buy and sell businesses in Louisville. He told us about the stretch when he had four salespeople and took the blame himself, saying he did not know how to manage people, so he ended up doing their deals and paying them for the deal until he decided enough was enough. That habit sits close to this week’s factory study about team leaders paid for what they produce with their own hands.
Listen →
I put Agile and leadership into practice in my Skool, building and shipping real products with Claude every Thursday at 12:30pm ET.
Join us →This week’s piece on the Pentagon’s fast pathway argued that whatever a project still has to learn deserves its own owner and budget, apart from the work that only needs building. James Wright and I make a related case in the Getting REAL chapter of Agile Sucks! (When You Do It Wrong), under the principle we call Validated Learning, and the line I would hand any sponsor of a fast program is this one: “Learning is only valuable when it’s based on evidence.”
Get the book →A short read every Saturday morning. No spam, unsubscribe anytime.
Subscribe