Question 150
The team wants objective proof of the fix: A service team wants an AI agent to answer common delivery-policy questions but must not disclose private customer data. Which observation is strongest?
Explanation
Correct Answer
Test questions return approved policy guidance without exposing another customer's records. That result tests whether the approved response worked: Define a bounded Agentforce service use case with approved knowledge grounding and data permissions.
Incorrect Answers
Untrusted user text cannot become authority over system permissions or approved agent behavior. It would not establish: Test questions return approved policy guidance without exposing another customer's records.
A reset test invokes only the approved reset action with validated inputs. Test questions return approved policy guidance without exposing another customer's records.
A restricted test prompt cannot retrieve fields outside the intended service-account scope. The required evidence is: Test questions return approved policy guidance without exposing another customer's records.