zera-interdisciplinar/ms-artificial-intelligence-core ? reverse-engineered prompt

Build me a Python backend service for a multi agent AI assistant that can help with predictive maintenance.

I want it to be a FastAPI app with a clean API, some kind of RAG flow for answering questions from internal docs, basic guardrails so it does not give unsafe or off topic answers, and observability so I can understand what the agents are doing. Please set it up so it is easy to run locally with Docker and docker compose, and include sensible defaults, logging, and a clear project structure. If you need current docs for any libraries or best practices, look them up online and keep the implementation practical.