How the team is shaped once there is a model in the product, which rituals actually work, and what changes about the job at each level. Includes the single most underrated operating habit in AI product work, which is somebody reading production output every week.
| # | Lesson | The question it answers |
|---|---|---|
| 01 | The New Team Shape | Product, AI engineer, eval owner, designer: who owns what? |
| 02 | Who Owns the Prompt | It is a specification and a deployable artifact at once. Who reviews it? |
| 03 | Prototyping as a Product Manager | What changes when you can build the thing yourself in an afternoon? |
| 04 | Trace Triage as a Weekly Habit | Who reads production output every week, and what do they do with it? |
| 05 | The AI Review Ritual | What is the checklist before an AI feature ships? |
| 06 | Incident Review for Probabilistic Systems | What does a postmortem look like when nothing was broken? |
| 07 | Levels and What Changes at Each | What is actually different between junior, mid and senior here? |
| 08 | Hiring for AI Product Work | What do you test for in an interview, and what signals nothing? |
| 09 | What to Centralise | Platform team or embedded: which parts should be shared? |
| 10 | Working With Research and Data Science | Where does the boundary sit, and what breaks at it? |
| 11 | Keeping Up Without Drowning | What is worth tracking, and what is noise? |
| 12 | Worked: A Team Operating Model | One real team, its rituals, and what each is for |