My goal is to develop a RAG-powered automation within MCP, where users can deploy custom AI models connected to their private or enterprise datasets. It would combine data retrieval, context awareness, and secure execution — turning MCP into a powerful environment for knowledge-driven automation.

Thanks @Simplita Ai — exactly! For embedding storage, I’m planning to use a vector database layer within MCP that securely indexes user data. Each deployment would have its own isolated namespace to ensure separation between users. For access control, I’m integrating token-based authentication and role-based permissions so that only authorized users or services can query embeddings or run model inferences. The goal is to make RAG workflows both scalable and privacy-safe within the MCP environment.