A startup founder is building an AI-powered compliance chatbot that answers employees' questions about internal security policies. When employees ask about a new policy that was added last week, the chatbot says it doesn't know about it. The chatbot uses an LLM with a fixed training cutoff date. What architectural fix addresses this gap?

• A.Increase the LLM's context window so it can hold all policies in its prompt at once
• B.Retrain the LLM on all company policies weekly to keep the model up to date
• C.Implement a RAG (Retrieval-Augmented Generation) pipeline that indexes the organization's policy documents and retrieves relevant sections at query time, allowing the chatbot to answer questions about newly added policies without retraining the base LLM
• D.Use a more recent LLM version with a later training cutoff date