In the previous post I talked about appetites.
In the Shape Up philosophy, once you set an appetite (e.g. six weeks), you place a bet against it. For example, in my last round, which started in August, I bet that I could build the recommendation engine in six weeks. Said another way, I gave myself six weeks to build a recommendation engine.
The bet was the whole round, not any single task inside it. In the round I completed 18 issues, and I never bet on one of them by itself. An issue belonged in the round only if it served the recommendation engine build. On Day 24 I had four issues to sort, and I asked one question of each: “Does this serve the recommendation engine?” Season-level ratings and actor pages did not, so they left the round.
Why is this important to TVLens? Well, this is why I will not be opening TVLens to the public right away. I built the engine before the interface. This decision was hard for me because I want to assemble everything up front. This is especially true because I built the TVLens recommendation engine with Claude. Building with AI requires restraint because it can build everything you want as long as you have enough credits.
I think building the recommendation engine was a good stopping point because it allowed me to not worry about the website, going public, and all the other things that plague the execution of a project.
All I have is a recommendation engine, and a pretty good one, I hope.