Libraries deploy a retrieval assistant over digitized local-history collections. Which use-case pattern matches?

Libraries deploying a retrieval assistant over digitized local-history collections match a knowledge-assistant / RAG-style municipal use case: index the document corpus, retrieve relevant material at query time, and answer patron questions grounded in those sources rather than inventing unsupported narratives. A fine-tuning workflow that retrains a language model on library catalog records is a model-training pattern that adjusts parameters rather than retrieving and grounding answers in the digitized local-history corpus patrons need to cite. An extractive summarization service that condenses finding aids for web display is a display-layer summarization tool, not a retrieval assistant that answers patron questions against the digitized local-history collections. An autonomous agent workflow that updates external catalog systems from chat is an action-taking integration that changes external state rather than grounding answers in the digitized collections patrons expect to cite. Exam caveat: RAG assistants still need corpus quality, access controls for sensitive or restricted records, citation behavior where required, and clear failure responses when retrieval returns nothing useful. Operational check: verify index coverage of the local-history collections, test representative patron queries for grounded answers with resolvable sources, refresh the index when new digitized material is added, and keep catalog-record fine-tuning or finding-aid summarization designs out of this knowledge-assistant path.