AI and agents
Retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is a method where an AI system first searches your own documents or data for relevant passages, then gives them to a language model to answer from.
How it works in practice
A language model only knows what it was trained on and what is in front of it. RAG puts the right part of your knowledge in front of it at the moment of the question. Documents are split into passages, each passage is turned into an embedding and stored, usually in a vector database, and a question retrieves the closest passages before the model writes its answer.
Most of the quality comes from the retrieval side: how documents are split, whether old versions are removed, whether the search also matches exact terms such as part numbers, and whether the answer cites its source. A RAG system answering from a stale handbook will be confidently wrong, so keeping the source documents current matters as much as the model.
A worked example
A 200-person manufacturer’s support team answers questions from 400 product manuals. A RAG assistant searches the manuals for each question, drafts an answer with links to the page it used, and the agent checks the link before replying to the customer.
Examples in this glossary describe typical builds for companies of 1 to 1,000 people. They are not named client results.















