Retrieval-Augmented Generation (RAG) is a technique specifically designed for this scenario. It enables an AI agent to connect to an external knowledge base, such as a repository of company policy documents. When a user asks a question, the RAG system first retrieves the most relevant document snippets and then provides them to the large language model as context. This ensures the generated response is accurate, up-to-date, and grounded in the company's official documentation, preventing factual inaccuracies or "hallucinations."

Static responses only: This approach is not dynamic and cannot generate answers from the content of external documents.

Manual lookup: This is the human process that the AI agent is intended to automate, not the method it uses.

Disable external data: This is the opposite of what is required, as the agent must access the policy documents.

Microsoft Learn, "Generative answers in Microsoft Copilot Studio". This document details how Copilot Studio uses a RAG pattern to generate answers from specified data sources. See the section: "Supported sources".

Microsoft Azure AI Documentation, "Retrieval-Augmented Generation (RAG) in Azure AI Studio". This provides a foundational overview of the RAG pattern. See the section: "What is Retrieval-Augmented Generation?".