pgvector for RAG: When you don’t need a dedicated vector database

Not every RAG stack needs another database. If your application already runs on PostgreSQL, pgvector may be enough for storing embeddings, filtering results, and running semantic search.