Vector Embeddings (with OpenAI and Supabase) - Part 3 (storing & searching)

Hey there! In the last two blogs, we talked about what vector embeddings are and how to set up OpenAI and Supabase locally so you have a solid playground to experiment in. So incase you want to check out the theory or have your set up up and running, go to the blogs below. Part 1 – The theory (what embeddings are and why they matter) Part 2 – Setup ( set up OpenAI client + Supabase + pgvector) Part 3 – Vector database & search (👈 YOU ARE HERE) Part 4 – A chatbot (proof of concept) This post is where we actually put everything to work: we’ll turn plain text into vectors, store them in a pgvector‑powered table, and run semantic searches over that data using a custom SQL function. From there, we’ll refactor the logic into small, reusable functions and add the Chat Completions API so the final experience feels like a friendly, grounded assistant instead of a raw database query. Also, it may look a little overwhelming for now, but trust me, if I can do it