What does an individual retriever do within an Agentforce-grounded prompt or action?
Select an answer to reveal the explanation.
Short Explanation
A retriever is basically the librarian for one shelf — point it at a data source, tell it how to search (keyword, vector, hybrid), and it fetches back the passages worth handing to the model.
Full Explanation
An individual retriever is configured against one specific data source — for example, Ridgeline's discharge-instructions library — with a chosen search type, and at run time it queries that source and returns the top relevant chunks to ground a prompt or agent action's response. It is the fundamental building block that either stands alone or gets combined into an ensemble retriever pulling from multiple sources. Retention or deletion of stale content is a data lifecycle/library management concern, not something a retriever itself performs. Speech-to-text conversion is a channel or input-processing capability entirely separate from grounding retrieval — retrievers work on text queries against indexed content, not audio. Document-level visibility is governed by Salesforce sharing and the Agent User's permissions, not by the retriever's search configuration; a retriever can only ever surface content the querying context is otherwise permitted to see. For the exam, know that a retriever is defined by its data source plus its search type, and it is the mechanism that actually executes retrieval at query time.