Language models know a lot about the world – but nothing about your company. Retrieval Augmented Generation, or RAG for short, closes this gap: before the model answers, the relevant passages are retrieved from your own documents and provided as context. The answer is thus based on your knowledge and can be verified at any time via the source reference.
What matters in practice
- Data quality: Outdated or contradictory documents lead to poor answers. Preparing the sources is the most important step.
- Access permissions: The assistant may only show what the person asking is allowed to see.
- Quality measurement: Test questions and ratings are used to continuously monitor answer quality.