Why Should a Fast Team's Memory Live in AI?
Shiva Rajaraman says fast-moving organizations have no memory layer unless everyone meets, so the memory layer belongs on the AI side. When a crew starts a set built around a huge tub they will fill so it looks like something else, AI should instantly return the three things to remember and warn that one step will not work.
A hundred people run into problems and solutions all day, and making the lessons available means the next idea comes back with what to watch out for. Knowledge that often sits in one person's head becomes available to everyone, so on every video and every production the errors that can make the work go wrong are no longer surprises. Someone who joins the team on day one should already have the history of everyone who has worked there for 10 years, and the team should keep refining the history as new lessons appear.
AI is a way to institutionalize capture of every nugget into one place, then vote each lesson up or down on whether it applies across everything or only in one situation. Once the base is built, the wisdom brings exponential returns, and that is the most important part of the work.
