1 article on experiment velocity.
After shipping 12 features in a quarter and moving zero meaningful metrics, I learned the hard way that AI products are not software projects. The roadmap is a hypothesis board, not a delivery schedule.
