Rewrite a vague question into a sharper RAG search query

Users ask short, messy questions, and my RAG system searches with exactly that messy wording.

Intermediate to advancedRead 5 minDoing: an afternoon, once you have a handful of your own real vague questions to test against6 steps

▶ Start

What you'll have at the end

A rewrite step that moves the real answer into first place on 5 real questions whose raw wording buried it third or worse, confirmed on your own retriever's scores.

Before you start

You need

A RAG pipeline that already embeds a question and retrieves against a vector index, plus a way to call an LLM before that retrieval step runs.

Not covered

Reordering or scoring chunks after they already come back from retrieval, and merging in a keyword index; this only changes the text sent into retrieval, before anything gets scored.

Leans on

6 steps

Where to go next

Add a reranker so RAG stops burying the right chunk

when the right chunk is already inside your retrieved batch but ranked below noisier ones, not missing because of the question's own wording

RAG's vector search misses exact terms that hybrid search catches

when the miss is an exact code or identifier vector search cannot see at all, not a vocabulary gap a rewrite can close

Your RAG's retrieval hit rate has never been measured

to measure this fix's real effect across your whole question set instead of the handful you spot-checked here

Checked 15 Aug 2026

Part of the RAG (Retrieval Augmented Generation) cookbook