RAG & Search
Retriever
The component that searches a knowledge source and returns relevant items.
Example
The retriever receives a query and selects likely source passages before the model writes an answer.
Why it matters
It narrows a large collection to material an assistant can realistically examine.
You'll hear it
“The answer writer never received the relevant passage from search.”
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
- Can a good answer-writing AI compensate for every search failure?
- No. It cannot reliably use a document that never reaches it.
- Can a retriever return too much material?
- Yes. A large set of weak matches can bury the useful evidence.
- Does it have to search the whole internet?
- No. It may search only a selected collection, such as internal documents.