You Probably Don't Need a Vector Database for Your RAG — Yet

“Vector Search is just matrix multiplication. And Python already has some of the world’s best tools for that.”

The vector database hype has convinced teams to add Pinecone or Weaviate to their stack before they’ve even validated their RAG prototype works. This piece makes the obvious-but-underappreciated point: NumPy can search millions of vectors in milliseconds, in memory, with no external dependencies. Save the specialized infrastructure for when you actually have hundreds of millions of vectors and need disk-based indexing. Most internal tools and MVP agents will never get there.