Retrieval-augmented generation (RAG) in plain English
2026-11-10
Definition
Retrieval-augmented generation, usually shortened to RAG, is a method where the system first searches your own documents for relevant passages and then hands those passages to an AI model so that it answers from them.
How it works
The steps are simple. A question comes in. The system finds the most relevant pieces of your handbook, price list or past emails. It places them in the model's context window along with the question. The model writes an answer based on that material and can point to where it came from. Because the answer rests on your text, hallucination drops and the knowledge stays current. When a document changes, you update the file and not the model.
RAG often suits small businesses better than fine-tuning, which changes the model itself and costs more to maintain. If you are weighing whether to build or buy, see build vs buy AI software.
A hypothetical example
Imagine a building supplies wholesaler with a 60-page returns and warranty policy. A RAG assistant lets the sales team ask "Can we refund a cut timber board after 14 days?" and see the exact clause it relied on.
If you want a system like this built around your own files, see AI development.