Glossary — Language models
RAG
RAG is a technique where relevant documents are retrieved before the model answers and handed to it alongside the question — so it answers from your data rather than from memory.
2 tokensRAG
Simply put#
The acronym is retrieval-augmented generation. First the passages in your documents that relate to the question are found, then they are given to the model together with the question.
Example#
An internal company assistant answering from your own procedures. The model never saw those procedures in training — they are supplied fresh every time.
Why it matters to you#
This is almost always the right first choice when you want AI answering from your data. Cheaper than fine-tuning, updatable instantly, and it reduces hallucination because the source travels with the question.