A pilot proves that a capability can work in selected conditions. Production readiness asks whether people can trust, operate, support, and improve it under real conditions.

Purpose and ownership

Is the business outcome and approved scope documented?
Who owns the process, the technical service, and the risk decision?
Who can pause or disable the workflow?

Data and knowledge

Which sources may the solution access?
Is the information current, authorised, and appropriately retained?
Can an output link back to evidence where decisions require traceability?

Permissions and actions

Which tools and records can the solution read or change?
Are permissions limited to what the use case requires?
Which actions require human approval?

Quality and evaluation

Has the system been tested against normal, difficult, ambiguous, and adversarial cases?
Are quality thresholds and escalation conditions defined?
Can failures and inappropriate outputs be identified after launch?

Operations and recovery

Are executions, errors, changes, and costs observable?
Is there a safe fallback when a model, connection, or data source is unavailable?
Do support teams have documentation and a recovery procedure?

People and change

Do users understand what the system does, what it does not do, and when they remain accountable?
Is feedback captured and reviewed?
Are process measures reviewed alongside technical measures?

Governance should be proportional to the impact of the use case. The goal is not paperwork; it is clear authority and evidence that allow useful AI to scale responsibly.