Vol.046 — Do You Know What a “Botsitter” Does?
Date: 2026-06-18 | Newsletter
Key Summary
In Vol.046, Zenta pushes back on a new label. Glean’s Work AI Institute — with researchers from Notre Dame, Stanford and Berkeley — surveyed 6,000 office workers and coined “botsitting”: the invisible labour of feeding AI context it should already have, checking its output, and cleaning up its mistakes, clocked at 6.4 hours a week. The data is a paradox. AI saves these same workers around 11 hours a week, then the context-feeding and cleanup quietly eat the savings back; 87% use AI at work and 75% feel more productive, yet only 13% say their company is actually performing better, and the heaviest botsitters are 73% more likely to be looking for the exit. Zenta splits the work in two: let the machine grade the mechanical half, and keep humans on the subjective stress-reads no agent can self-grade — that is judgment, not babysitting. He adds the Asian “harmony tax” angle — correcting an AI costs no face — and dares readers to rename the role themselves. Do it simple. Let the machine grade the machine, and keep your humans on the work only a human can feel.