Minimal RAG system architecture integrating a web/mobile frontend with a containerized backend that orchestrates PostgreSQL, vector storage, and Mistral LLM for semantic search and generation. User requests flow from the frontend through the backend Docker container, which queries both the relational database and vector store, then sends context to the Mistral LLM server for augmented responses. This three-tier pattern isolates presentation, application logic, and data/AI layers, enabling independent scaling and technology swaps. Fork this diagram on Diagrams.so to customize your LLM provider, vector database, or containerization strategy. The bidirectional connection between Mistral and vector storage highlights the retrieval-in-the-loop pattern central to production RAG systems.