hyperfiddle

2025-10-13T22:30:34.364429Z

that would be cool

Dustin Getz (Hyperfiddle) 2025-10-12T13:45:46.762629Z

Yeah the main issue here is that these AI generations accumulate and get large, so it seems to me like you want to use a virtual scroll pattern for this so that the state is on the server, and durable, you can refresh the page and get it back, and it doesnt matter how large your history is

Dustin Getz (Hyperfiddle) 2025-10-12T13:48:29.590959Z

in other words, use clojure to interact with the model provider and accumulate the result into a clojure data structure, and then use Electric to make a UI that attaches to it and windows over it

2025-10-11T12:00:07.418069Z

I really wonder if anthropic didn't secretly train on my api inputs of last year anyway 😆, Struggled quite a bit with missionary + llm streams using 3.5. Thanks for the snippet.

👀 1
😂 1
2025-10-11T05:19:08.498799Z

https://gist.github.com/tommy-mor/7fcd3eeed28ea0c9fb5dbb864b048043 4.5 sonnet spit this out, and it worked first try!

👀 1