Model Context Protocol, or MCP, is an open-source standard for connecting AI applications to external data and tools. It gives an application a shared way to discover what a connected server offers, request information and call functions. MCP standardizes communication with connected servers. The AI application uses its own language model. Its presence does not tell you which information a particular connector can read or which actions it can perform.
Retrieval-augmented generation, or RAG, combines a language model with a system that retrieves relevant text before the model writes an answer. Those passages can supply information beyond the model's training, but the model still generates the response rather than merely returning a search result. The original RAG paper was first submitted on May 22, 2020. This is an explainer of an established research approach, not a November 2026 announcement. Original paper record