Retrieval
Retrieval is finding relevant information from an external collection in response to a query. Its output can be documents or passages that an application supplies to a model.
[LangChain]In practice · hypothetical example
A help desk searches product manuals and returns the passages most relevant to a customer’s error message.
[LangChain]A little deeper
Retrieval is a search step. Combining its results with generated answers creates a broader system; the retrieved passages alone are not that generated answer. [LangChain]
A common mix-up
Retrieval and generating an answer are the same operation.
Retrieval returns material; generation uses input to produce a response. [LangChain]
A system returns matching manual passages without writing an answer. What did it do?
Sources & editorial notes
Evidence: supported. Primary-source support for this scoped entry; publication approved by the project owner.
- Retrieval ↗ (opens in new tab)LangChain · Publication date unknown
Relevant section: From retrieval to RAG; Retrieval pipeline
Last editorial review: 2026-09-13 by project-owner.
First observed in this corpus: Unknown.
Revision history
Revision 2 · Created 2026-09-13 · Updated 2026-09-13
Project owner approved the current content for publication. Existing evidence scope and limitations remain applicable.