Implementation risk questions
A capability-led review should examine the conditions that could make an autonomous workflow unreliable, unaccountable, or difficult to operate.
Foundation readiness
Check whether the data, lineage, semantics, and decision rights needed by the workflow are explicit and available. Where they are not, record the missing foundation work before expanding the agent boundary.
Trust governance
Decide how actions will be observed and evaluated, who can challenge them, and how an authorized person can pause or reverse them. Treat those controls as part of the scoped capability.
Role and ownership gaps
Identify the people accountable for policy, exceptions, evaluation, and recourse. Make changes to roles and operating routines explicit rather than leaving them implied by the technology design.
Platform dependency
Separate platform-specific choices from the enterprise definitions they support. Data contracts, decision policy, workflow boundaries, and governance should remain inspectable across the operating layer.
Scope and expectation alignment
Describe the operating domain, approved agent boundary, unresolved constraints, and evidence required for extension. This gives sponsors and operators a shared basis for reviewing the capability.