Today
Every new part starts its FMEA almost from scratch, in long workshops. Failures that already happened on similar parts, in the field or at another plant, never reach the analysis.
An example
Arnstein is developing a new EPDM seal profile DP-0870. The agent pulls 14 failure modes from three predecessor FMEAs and two field complaints and points out that one occurrence rating is lower than the predecessors' complaint rate suggests.
Your data, wherever it lives
Today earlier FMEAs sit in the CAQ system or as spreadsheets on drives, 8D and test reports as PDFs in the QMS, complaints in the CAQ system and BOMs in the PLM and ERP. demi reads, understands and links them per part and product family, with your rights. Before the first workshop, the AI lays out a draft with failure modes, causes and actions and their evidence, so no FMEA starts from scratch again, and your team rates and releases.
Where the data lives today
- QM system
- Office files
- Folders & drives
- PDFs
- PLM
- ERP
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
The case needs data from the PLM, QMS, CAQ, ERP and spreadsheets. demi replaces nothing and links existing FMEAs, 8D reports, complaints, test reports and BOMs to the part and its product family. demi writes back only after approval.
- 2 · PermitPermission Layer
Each team sees the analyses of its own product lines. Ratings and release stay with the moderator and the responsible engineer, and every change is recorded.
- 3 · ActIntelligence Layer
The agent finds the closest predecessor parts and lists their failure modes with the recorded complaints and 8D cases behind them, each with its source. It drafts causes and actions and flags where an occurrence rating contradicts field data, but never sets a rating itself.
How it runs
- 1
Create the new part with its product family and material
- 2
Find similar predecessor parts and their FMEAs
- 3
Propose failure modes, causes and actions with evidence
- 4
Rate, add or reject them in the workshop
- 5
Release the FMEA; every change is recorded
What changes
Workshops start from a solid draft
Field experience flows back into new designs
Consistent analyses across plants
What does it bring you?
Check it with your own numbers
Every starting value is an assumption you can change. The calculation sits right below, step by step.
Your numbers
Result
+115 hfreed per year
Assumptions and calculation
demi proposes failure modes, causes and actions from past FMEAs and 8D reports; you rate them.
8 h → 5.6 h per case
How we calculate
4 FMEAs × 8 h × 30% × 1.0 (manual) × 12 months = 115 h
Value per year, before demi costs
Freed per year+115 h
Assumption: labor cost of $48.62 per hour in US manufacturing in June 2026 (BLS), plus workplace, IT and overhead.
= Value per year, before demi costs
+$7,475
gained per year
115 h ÷ 1,760 h per role = 0.1 full-time roles
An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.
Combine with more cases and get the packageQuestions about this use case
Who decides the severity, occurrence and detection ratings?
Your team, as before. The agent proposes failure modes, causes and actions and shows field data that calls a rating into question. A person always sets the values in the FMEA, and every change is recorded.
Can it read our existing FMEA exports and spreadsheets?
Yes. demi reads exports from your FMEA software as well as spreadsheets and maps lines, failure modes and actions to the part. You do not have to re-enter old analyses.
How does demi find similar parts for the FMEA?
Through product family, material, function and BOM structure, plus a comparison of the descriptions by meaning. For every proposed predecessor you see why it counts as similar.
How do I know where a proposal came from?
Every proposed failure mode links to the FMEA line, complaint or 8D report it came from. In the workshop, the team checks a source, not a claim.
