You'll see "RAG" mentioned constantly around AI products that let you "chat with your documents." It stands for retrieval-augmented generation, and the idea behind it is simpler than the acronym suggests.
- RAG retrieves relevant passages first, then hands them to the model along with your question.
- It lets a model answer accurately about content it was never trained on, without retraining.
- Answer quality depends heavily on the retrieval step finding the right passages.
The core idea
Instead of relying purely on what a model learned during training, a RAG system first searches a knowledge base (your documents, a database, the web) for relevant passages, then hands those passages to the model along with your question. The model answers using that retrieved material rather than guessing from memory.
Why it matters
It lets an AI tool answer accurately about content it was never trained on — your company's internal docs, yesterday's news, a PDF you just uploaded — without the expensive process of retraining the model itself.
The tradeoff to know about
A RAG system is only as good as its retrieval step — if it fetches the wrong passages, the model will confidently answer from the wrong material. Answer quality depends heavily on how well the underlying search is tuned.
Frequently Asked Questions
What does RAG stand for?
Retrieval-augmented generation — a technique where a model retrieves relevant passages from a knowledge base before answering, rather than relying only on what it learned during training.
Why use RAG instead of fine-tuning a model on your data?
RAG lets you add or update source content instantly without retraining anything — it's far cheaper and faster to keep current than fine-tuning.