I spend most of my days building and distributing AI-powered products. Not one product — a few, in parallel, through a studio model. The goal is simple: short feedback loops, a small team, and a rapid cadence.
Lean by default
Every decision is a tradeoff between speed and scope. A lean team forces the right defaults: own the distribution, write the copy yourself, ship the prototype before the plan. If something needs five people to move, it is probably the wrong thing to build.
Distribution is a first-class product decision, not an afterthought.
Cadence over perfection
The cadence is the moat. Weekly releases, daily writing, constant contact with users. Perfection is expensive; learning is cheap. I would rather ship something rough and learn from real users than polish something that never meets them.
More on the specific mechanics — how we pick bets, how we kill them, how we staff them — in future posts.