I built a Q&A app for AI-powered search over Naval's Almanack (https://www.navalmanack.com/almanack-of-naval-ravikant/table-of-contents)
Here's a demo of the app: https://twitter.com/varin_nair/status/1644384605133209627
I used Supabase DB with pgvector to store my embeddings, and used a Supabase function to find the essay chunk embeddings that were most similar to the user query embedding.