Retrieval-augmented generation: relevant passages are first pulled from a trusted document collection, and the model writes its answer from them.
Retrieval-Augmented Generation (RAG) gives a Large Language Model (LLM) the documents it needs at the moment it answers, so the answer rests on sources the organization trusts.
Instead of relying solely on the static knowledge the model was originally trained on, a RAG system intercepts a user's prompt and performs a real-time search against a secure corporate vector database. It retrieves the most relevant proprietary documents and feeds them to the LLM as context.
This grounds the AI's answer in verified, up-to-date enterprise data, reducing hallucinations while respecting internal data governance and access controls.
See also Vector Database, Data Governance, Data Mobilization and Data Lakehouse.