AI readiness self-assessment

How clean and accessible is the data an agent would act on? weight 25

Exploring: Scattered, unclear ownership Piloting: Central but inconsistent Governed: Governed with owners and quality checks Scaled: Governed and exposed through APIs

Is there a policy for what AI may do without a human? weight 20

Exploring: No Piloting: Drafted Governed: Approved and communicated Scaled: Enforced in systems with audit trail

How are approval thresholds defined? weight 15

Exploring: Informally Piloting: Documented Governed: Configured in a system Scaled: Configured with escalation and review of exceptions

Who can evaluate an AI vendor's claims? weight 15

Exploring: Nobody in particular Piloting: One enthusiast Governed: A named team with a checklist Scaled: A team with a checklist and vendor test results

How is data shared with an AI vendor controlled? weight 15

Exploring: Not controlled Piloting: Reviewed per case Governed: Standard DPA and review Scaled: DPA, review, logging and periodic audit

How would you know whether an agent helped? weight 10

Exploring: We would not Piloting: Anecdotes Governed: Defined KPIs per use case Scaled: KPIs measured against a baseline