Glossary
Retrieval-augmented generation is a pattern where a system searches your own documents for relevant passages and gives them to a language model as context, so the answer is grounded in your material rather than the model's memory.
A question arrives; the system retrieves the passages most relevant to it from a body of content you control; those passages are supplied to the model along with the question; the model composes an answer from what it was given.
It keeps answers tied to source material you can point at, it updates the moment the underlying documents change, and it makes citation possible: the system can show which passage an answer came from.
If the right passage is not found, no model will produce a correct answer from what it did receive. Most disappointing RAG systems are retrieval problems wearing an AI costume: badly chunked documents, missing content, or a corpus nobody curated.
A system that answers everything will answer wrongly at the edges. A system that escalates to a person when retrieval is weak, carrying the full context with it, is the one businesses can actually stand behind.
For factual, changing business content, usually yes. RAG updates when the documents update and can cite its source. Fine-tuning teaches style and format well but bakes facts into the model, where they go stale and cannot be traced.
Most often the retrieval step failed rather than the model. Check whether the correct passage exists in the corpus at all, whether it was chunked so the answer survives intact, and whether the system escalates instead of guessing when confidence is low.
Retrieval-augmented generation is a pattern where a system searches your own documents for relevant passages and gives them to a language model as context, so the answer is grounded in your material rather than the model's memory.
