@neumann @nate I loved this episode. Your ability to describe coding and calling external APIs that are slow, scattered and nested, through narration is uncanny and awesome. 🙂 I found myself constantly nodding as you were describing making calls to the database and the MAM — especially when you were describing having to reassemble denormalized data, and caching them locally. My example of having to cache data was using OpenAI’s API for a huge batch operation, as well as calling some sort of readability score generation — in the latter case, I cached it wasn’t because of cost, but it was so computationally expensive. I calculated the readability score of 100s of commits of my book manuscript, and it took tens of seconds per call. I remember generating all the scores using pmap, which took tens of minutes, and then caching them away so I never needed to do that computation again. Keep up the great work!