# Not Every RAG System Needs a Vector Database

Everyone building RAG systems starts the same way. Document → Chunks → Embeddings → Vector Database → Similarity Search → LLM That pipeline works. But it is not the only way to retrieve information. A