T3 chat cloneathon suggestions
I'm trying to clone t3 chat, just to understand the level of optimisation and stuff Theo did to make it feel faster than other LLM interfaces... Here are a few things I noticed or grasped...
1. Local first approach -> to make every navigation quicker...
2. Stream sync -> to make stream resumable on reloads and etc ....
3. Storing stuff in IndexedDB for faster execution and syncing it with the actual database...
I want to know, am I on the right path ?? I have made an initial MVP, without any optimisations yet... If anyone can give more insights on the local first approach and indexDB stuff, it would be helpful.... Why do these things matter ?? And how can I sync, LLM response, indexDB and actual remote database to avoid inconsistency ...
(Not actually interested in cloneathon, just need learning resources to understand stuff better) ...
1 Reply
you're having the same question as me about the syncing layer. I'll also wait for the answer!