
Trust, Review, and Work Restructuring with AI

How to measure AI developer productivity in 2025 | Nicole Forsgren
Topic 6 of 18
Published Oct 19, 2025 · Clip 17:41–21:21 (3:40)
Echoing past talks, Nicole reaffirms trust in AI-generated code as a major challenge, with developers spending more time reviewing than writing, evaluating for hallucinations, reliability, and style. This shifts attention dynamics, breaking traditional flow but enabling shorter blocks via machine-assisted context recovery and diagrams. She speculates on restructuring work around 4-hour deep work limits (per Gloria Mark), making 45-minute blocks viable for AI-assisted ramp-up, turning engineers into coordinators of AI agents.