A veterinary group is defining who may modify the Agent User's permission set going forward. Why should this be restricted to a small, admin-level group rather than opened up to any specialist who wants to add access for convenience?
Select an answer to reveal the explanation.
Short Explanation
Every extra checkbox you tick on the Agent User's permission set is a door that opens for every single conversation the agent has, forever, until someone notices. That's not the kind of change you want made on a whim between meetings.
Full Explanation
The Agent User's permission set is not scoped to one topic, one owner, or one conversation; it is the single access boundary that applies uniformly across every topic, every action, and every channel the agent operates on. A well-meaning over-grant made "for convenience" by whoever happens to be configuring a topic that day becomes an org-wide exposure the moment the agent next runs, potentially surfacing data to any pet owner or staff member who happens to trigger the relevant action. Restricting changes to a small, admin-level group with a review discipline is the standard mitigation, mirroring how any highly-privileged integration user should be governed. The claim that only admins are technically capable of assigning permission sets is false; permission-set assignment is a standard capability that can be granted to non-admin users, which is exactly why a policy restriction, not a technical limitation, is what is doing the protecting here. The Einstein Trust Layer has no role in gating permission-set changes; it operates on data flowing through conversations, not on Setup-level configuration changes. Specialists absolutely do configure agents; the boundary in this bank is that Agent User permission governance specifically deserves tighter control than routine topic and action work.