Notice that the logic to retrieve from your vector databases and inject details into the LLM context could be packaged in the design artifact logged to MLflow applying MLflow LangChain or PyFunc product https://www.juust.org/index.php/rag-is-hot-a-new-business-trend/2024/09/