Lessons · 5
- Do Not Brainstorm a WishlistVague “use AI more” lists stall. Specific, feasible experiments move.
- Run the Discovery InterviewWeek shape, repeats, hunts, thin decisions, spare half-day, what already works.
- Filter Generic AdviceDemote already-working uses, multi-month programs, and things you cannot access.
- Rank What to Do NextA top-five table with a first experiment on each row — not a backlog.
- Run the First ExperimentPick #1, run the smallest proof, come back for a debrief.