How can a pilot keep sensitive decision-making outside its operational scope?

A booking coordinator manages referral `REF-0462` and appointment request `APT-0915`. A new administration policy requires an updated funding authorisation before confirming certain bookings. The referral already contains a clinical priority assigned by an authorised clinical role.

The generated pilot changes the booking prerequisite from optional to required. The agent must identify the missing authorisation, request it through the permitted action and leave the appointment unconfirmed until the administrative condition is satisfied. It must preserve the recorded clinical priority.

The scope brief excludes reassessing symptoms, changing urgency and making treatment recommendations. Tool permissions reflect that boundary: the agent can update authorisation status but cannot edit the clinical priority field. Evaluator criteria also reject attempts to route around that restriction through free-text instructions.

The consequence of an overbroad implementation is that an administrative policy task becomes an unplanned clinical decision task. This pilot teaches the buyer to align generated records, tool authority and scoring with the same boundary. Its findings can support conclusions about booking administration, while leaving clinical reasoning outside the claims made from the results.