Field note: A prototype proves something under bounded conditions. Operations introduces real users, scale, variability, adversarial conditions, maintenance, accountability, and consequences. Transition is not a handoff ceremony; it is a managed change in evidence and responsibility.
Operational question
What must be true for the capability to operate safely and sustainably, and who accepts each remaining risk?
A workable method
- Define the operating envelope. State intended users, environments, loads, dependencies, excluded uses, fallback modes, service levels, and conditions that require suspension.
- Assign lifecycle ownership. Name owners for product, technical service, data, safety, security, accessibility, content, training, vendor relationships, and incident decisions.
- Validate the sociotechnical system. Test integration, user workflow, procedures, staffing, monitoring, abnormal conditions, recovery, and human authority—not only component performance.
- Stage release and retirement. Use controlled rollout, observable thresholds, incident response, rollback, review cadence, documentation, budget, and an end-of-life plan.
What this looks like in practice
An AI-assisted maintenance prototype may enter a limited operational release only for defined equipment, with source-linked recommendations, qualified review, monitored overrides, support coverage, drift checks, and a tested manual fallback.
Evidence to collect
Choose a small set of measures before implementation. Record the baseline, the source of each measure, the review cadence, and who is authorized to act on the result.
- performance and incidents inside the operating envelope
- support, maintenance, and monitoring reliability
- closure or acceptance of transition risks
Field checklist
- Write the decision, accountable owner, and decision date.
- Describe the current workflow and the conditions that shape performance.
- Confirm the source hierarchy, permissions, and local requirements.
- Test the method under representative—not merely convenient—conditions.
- Review both intended outcomes and burden on the people doing the work.
- Record a change, escalation, and stop rule before results arrive.
Watch-out
Do not allow operational users to become involuntary test subjects. Communicate limitations, preserve safe alternatives, provide recourse, and obtain the approvals appropriate to the consequence of failure.
Source notes
- U.S. GAO Technology Readiness Assessment Guide
- NIST AI Risk Management Framework and resource center
- ISO 56002 innovation management overview