TL;DR:
- chunk files, index chunks
- vector/hybrid search over the index
- node app to handle requests (was the quickest to implement, LLMs understand OpenAPI well)
I'm not too much into React but this seems like the kind of tools that can massively increase productivity. I can see open-source being a good source of both adoption and virtuousness as well. Very cool!
TL;DR: - chunk files, index chunks - vector/hybrid search over the index - node app to handle requests (was the quickest to implement, LLMs understand OpenAPI well)
I wrote about it here: https://laurentcazanove.com/blog/obsidian-rag-api