A medical device contract manufacturer managed non-conformances on paper and spreadsheets, and many stayed open for weeks. A digital NCR workflow with AI-drafted descriptions and suggested similar cases cut average closure time from 34 to 15 days and overdue NCRs from 41 to 6.
Key numbers
A medical device contract manufacturer recorded non-conformances on paper forms, then transcribed them into a spreadsheet. Raising an NCR took time, details were often incomplete, and many NCRs sat open for weeks waiting for information, investigation or sign-off. In a regulated industry, a backlog of open NCRs is a compliance risk as well as an operational one.
The situation
Average time to close an NCR was 34 days. Forty-one NCRs were overdue. Quality engineers spent about 3.8 hours on each NCR, much of it gathering information, chasing people and writing up the same details in different formats.
What changed
The company replaced the paper system with a digital workflow, using AI in the way described in AI agents in manufacturing, drafting and suggesting, with people deciding:
- NCRs raised on tablets at the point of detection, with photos, part numbers and batch details captured immediately.
- Automatic routing to the right engineer, with reminders before due dates and escalation when actions became overdue.
- AI drafts the problem description from the operator's notes and photos, and suggests similar past NCRs with their root causes and corrective actions.
- Engineers confirm root causes and actions themselves. The AI's suggestions are a starting point, not a conclusion, keeping the investigation grounded in evidence as in a problem-solving culture with 8D.
- A dashboard of open NCRs by age and owner, reviewed at the weekly quality meeting.
Results
Average closure time fell from 34 days to 15 within six months of the new workflow going live. Overdue NCRs dropped from 41 to 6. Engineers' time per NCR fell from 3.8 to 1.6 hours, because information arrived complete and drafting took minutes instead of an hour.
Why it worked
Most of the delay in the old process was waiting and rework: incomplete information, NCRs sitting unnoticed and repeated writing. The digital workflow removed the waiting; the AI removed much of the writing; and suggested similar NCRs helped engineers find root causes faster. Throughout, accountability for conclusions stayed with qualified people, which matters in a regulated environment.
Lessons for your operation
- Capture NCRs at the point of detection, with photos, while details are fresh.
- Automate routing, reminders and escalation.
- Use AI to draft and to find similar past cases, not to decide root causes.
- Make open NCRs visible by age and owner.
- Keep qualified people accountable for every conclusion.
