How an enterprise RAG works, the architecture you need (Qdrant + n8n + LLM), real SMB use cases and how much it really costs to build an AI assistant that knows your company's information.