Ragbag

Ragbag is a starting point for chat-with-your-documents products. Users upload PDFs, Word files and web pages, which are chunked, embedded and stored in Postgres with pgvector. Chat answers stream through the Vercel AI SDK with inline citations that link back to the source page. It includes teams with shared libraries, Stripe metered billing on tokens, and an admin view of costs per account.

It is opinionated about Postgres for everything, vectors included, so there is one database to back up. Model providers are swappable through a single config file. The limitation: ingestion runs in the request cycle with a background fallback, and documents over a few hundred pages time out on serverless hosts unless you move parsing to a worker.