Start with real work
Choose a clearly bounded process and describe the expected result. Identify who performs it today, what information they need and how successful completion is recognized. This context guides the AI evaluation.
People and spaces
Define the participating teams and the contexts they need to share. Separate administration from day-to-day activity, and describe which information belongs in each workspace.
Tools and decisions
List the tools needed and the actions that could affect external systems. Determine which require human review, who can approve them and what evidence is needed before execution.
Visibility into what happened
A useful evaluation makes it possible to reconstruct the intent, tools used and outcome of an execution. That visibility helps review behavior and improve the configuration.
Agree on the scope
The evaluation document should distinguish included capabilities, limits, technical environment and acceptance criteria. Pricing, licensing, support and deployment terms are agreed for the specific proposal.