RAG (retrieval-augmented generation)
What is RAG (retrieval-augmented generation)?
A setup where the model looks things up in your own documents or database before it answers, instead of relying only on what it learned in training. This grounds replies in your data, cuts hallucinations and lets the model point to its sources. It's the most common way to put a company's own knowledge behind a chatbot.
Why it matters
A standard language model only knows what was in its training data, so it can go stale or invent details. RAG lets a business point the model at its own current information without retraining anything. For teams, that means answers built on what’s actually true inside the business right now: today’s pricing, the policy as it stands, the specific customer’s history. Every answer comes with sources you can check, which is what makes RAG the practical way to get trustworthy answers from proprietary data.
In practice
A support team connects RAG to its help-centre articles and past tickets. When an agent asks “how do we handle a refund after 30 days?”, the system pulls the relevant policy and drafts a reply grounded in it, with links to the source articles. Nothing is retrained; when the policy changes, the next answer reflects it automatically.
