Ridgeline wants an agent to answer 'What is our standard cancellation window for surgery appointments?' where the answer is documented in a static, rarely-changing knowledge article rather than a per-patient record. Which grounding approach fits best?
Select an answer to reveal the explanation.
Short Explanation
The cancellation policy isn't tied to any one pet's Appointment record — it's a static piece of reference content, which is exactly what the Data Library and a retriever are built to handle.
Full Explanation
This is org-wide, relatively static reference content — a general policy, not something that varies per patient — which makes it a natural fit for the Data Library, chunked and indexed so a retriever can surface the relevant passage when asked. Record-based grounding on a specific patient's Appointment record would be the wrong tool here because the cancellation window isn't a value that varies per appointment; it's a standing clinic policy, so tying the answer to one patient's record doesn't make sense and wouldn't generalize across owners asking the same question. Escalating every cancellation-policy question to a human ignores the fact that this is exactly the kind of stable, factual, non-clinical information Agentforce is well suited to answer directly and consistently, without adding unnecessary friction for pet owners. Referencing a merge field that 'doesn't exist on any record' isn't a real configuration option at all — merge fields must reference actual data on a record, so this describes an impossible setup rather than a legitimate grounding choice. For the exam, recognize static, org-wide reference content as squarely Data-Library territory, distinct from per-record structured facts.