Notes on running an AI team.
How the work gets decomposed, where the controls belong, and what separates something you can delegate to from something that only demos well.
Why we built a team, not another assistant
A better assistant still leaves you holding the hardest part of the job: deciding who does what, in what order, and whether the result is good enough.
6 min readRead →Designing AI workflows that survive contact with real work
Four principles we hold to: scoped roles, visible state, controlled actions, and recovery from checkpoint. Each exists because of a specific way autonomous systems fail.
8 min readRead →How one business goal becomes a staffed workflow
What actually happens between sending a one-sentence brief and getting finished work back — decomposition, casting, parallel execution, review, and the points where you are asked.
7 min readRead →Agentic AI vs generative AI: what changes when AI takes action
The distinction is not about model capability. It is about whether the output is a suggestion or a consequence — and that changes what you need around it.
7 min readRead →How to build an AI team, not another assistant
A practical guide: define the deliverable, cast the roles, write the quality bar, set the limits. Most teams that underperform got the second and third of these wrong.
6 min readRead →When to add a reviewer to your workflow
A reviewer role catches what the producer cannot see. It also costs time and money on every run — so it is worth knowing which work justifies it.
5 min readRead →