How should a model lab scope its first handover pilot?

Start with the decision the lab wants a model to handle better. 'Understand enterprise workflows' is too broad to guide a useful sample. 'Decide whether a receiving team can accept a case or must return it for missing prerequisites' provides a concrete boundary.

Then define the sending role, receiving role, entry point and stopping point. A collections handover might begin after a repayment arrangement fails and end when the receiving team accepts the case or records a justified return. Earlier account origination and later recovery activity can remain outside that pilot.

Ask MirrorFirm to scope generated records and tasks around that boundary, including cases that exercise the chosen distinction. Define the evidence the buyer will inspect and the acceptance criteria before expanding the domain or adding more tools.

The initial result should answer a specific question about usefulness: whether the material supports the targeted behaviour under the agreed conditions. It should not be treated as evidence that a model can manage the entire business process. A useful first step is a short scope brief describing the decision, available inputs and required output.