RAG & Search
Retrieval-augmented generation
RAG
AI looks up relevant information before answering you.
How it works
- AskYou ask a question.
- SearchIt searches a source.
- RetrieveIt finds the passage.
- AnswerIt answers using the passage.
Example
Your company connects an AI assistant to its HR handbook. When you ask about parental leave, it retrieves the relevant policy and uses it to answer.
Why it matters
It lets AI answer using information that wasn’t necessarily included in its original training.
You'll hear it
“We’re using RAG so the chatbot can answer questions about our internal documents.”
Can you control it?
Set by the team
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- What can RAG get wrong?
- It may find outdated or irrelevant material, miss a useful document, or misunderstand what it reads.
- Does RAG retrain the model?
- No. Looking up information for a reply does not itself change the AI's underlying learning.
- How can I judge a RAG answer?
- Open the cited sources, confirm they support each important claim, and notice whether relevant documents appear to be missing.