RAG (retrieval-augmented generation)
A method in which a language model first finds the relevant passages in your documents and bases its answer on them.
Glossary · Automation & AI
RAG (retrieval-augmented generation) combines search with a language model. When a question is asked, the system first finds the relevant passages in your documents, and the model then writes an answer based on exactly those.
This lets a model answer about things that were not in its training data, such as internal rules, product descriptions or contracts, and say which document the information came from. It lowers the chance of invented answers, though it does not remove it.
The quality of the result depends above all on the documents: if they are out of date, contradict each other or simply lack the information, the model will not fix that. A RAG project therefore often begins with putting the documents in order.