Add content to the
knowledge base

Upload documents or crawl a website. Content is chunked, embedded, and stored in pgvector. Re-ingesting an unchanged source is skipped automatically.