Spreadsheets you can edit
The ROI model is a working file with visible formulas, not a PDF of a chart.
Illustrative · pack available on request during pre-launch
Tools toevaluate thisproperly.
Material for the people who have to build the business case, run the security review and defend the decision internally — written to be useful even if you never buy anything.
What is available
Request any of these directly — there is no download-form maze.
Model
First-pass yield ROI model
A spreadsheet that turns your unit volume, scrap cost, rework hours and warranty rate into a defensible payback estimate.
GuideInstrumentation readiness guide
What telemetry a line needs before autonomy is worth attempting, and what to add if it is missing.
ChecklistPhysical-AI vendor evaluation checklist
The questions to ask any vendor claiming autonomy, including the ones that are awkward for us.
GuideShadow-mode baseline playbook
How to run an honest fourteen-day baseline and interpret the accuracy gap without fooling yourself.
PackIT/OT security review pack
Architecture, data-flow diagrams, isolation model and audit design for your security questionnaire.
ReferenceAppliance defect taxonomy
A working taxonomy of defects across forming, enamel, foam, assembly and test, with detection notes.
How the ROI model works
Most vendor ROI calculators produce a large number by multiplying an optimistic percentage against your revenue. Ours starts from unit economics you can verify from your own reporting.
Inputs: annual unit volume, first-pass yield, scrap cost per unit, rework hours and hourly rate, downgrade rate and margin delta, warranty claim rate and cost per claim.
Outputs: annual recoverable cost, modelled improvement range with explicit assumptions, subscription cost and payback period — with every assumption editable.
Deliberate design choices
- Conservative, midpoint and optimistic cases, all shown
- Assumptions on the same sheet as the results
- Subscription cost included, not hidden
- Ramp period modelled — value does not start on day one
- Works even if you conclude the answer is no
Resources map to the build
Every guide is anchored to a stage of the line, because the questions a foaming engineer asks are not the questions a test engineer asks.
Form & stampWeld & seamEnamel & paintFoam & insulateWire & assembleLeak & function testRate & pack
Reference ranges we work with
Indicative industry ranges used for modelling. These are estimates for planning, not measured Duromex results.
| Loss type | Typical range | Where it is created | Where it is found |
|---|---|---|---|
| First-pass yield shortfall | 3–12% of units | Across all stages | End-of-line test |
| Enamel downgrade | 1–5% of coated parts | Enamel and cure | Surface inspection |
| Foam void / density fault | 0.5–3% of cabinets | Foaming station | Energy or thermal test |
| Assembly / wiring rework | 1–6% of units | Assembly and harness | Functional test |
| Leak escape | 0.1–1% of units | Brazing, sealing, gasketing | Leak test or the field |
| Warranty claims | 1–4% within 24 months | Any upstream stage | The customer's home |
Ranges are drawn from public industry literature and partner conversations. Your plant's numbers are the only ones that matter for a business case.
Five steps to an internal business case
The sequence that has worked for the plants we have talked to.
Pick one loss
Choose the single loss with the clearest cost and the clearest owner.
Cost it honestly
Scrap, rework hours, downgrade margin and warranty — from your own reporting, not from a vendor.
Model the recoverable share
Use the conservative case. If it does not work conservatively, it does not work.
Define the exit
Write down what result would make you stop, not just what would make you continue.
Set the gate
Agree the baseline window and the measurement method before anything is installed.
Questions worth asking any vendor
Including us. If a vendor cannot answer these, that is your answer.
Baseline
How is the baseline measured, and who defines it?
Attribution
Can the system attribute an end-of-line failure to the upstream stage that caused it?
Latency
What is the station decision latency, measured, on real hardware?
Authority
Exactly what can the system write, within what bounds, and how is that revoked?
Offline
What happens to production when the WAN link drops?
Evidence
Can you export a complete evidence chain for a single serial number?
IP
Where does our recipe and defect data live, and who else's model does it train?
How resources are delivered
Practical, not promotional.
Written guides
Plain documents, no gating, sent on request within one business day.
Security pack
Architecture and data-flow material formatted for an IT/OT questionnaire.
Working sessions
For serious evaluations, a process engineer will work through your numbers with you.
How we treat resources
0
email walls before a download
1
business day to receive a pack
100%
assumptions shown in the ROI model
3
scenario cases in every model
What evaluators tell us
“Our end-of-line functional testers told us what failed. They never told us which press stroke or foam shot caused it. That link is the whole product.”
“Foam voids are invisible until the energy rating comes back wrong. Predicting density distribution before the cabinet cures is the part I could not buy anywhere else.”
“If it writes to the press, it needs an audit trail a quality auditor accepts. Duromex started there instead of bolting it on.”
About this material
Why is nothing downloadable yet?
During pre-launch we send resources directly so we can answer questions alongside them. Self-serve downloads arrive with general availability.
Can we use the ROI model without talking to sales?
Yes. Ask for it and we will send it. If you conclude the case does not work, that is a legitimate and useful outcome.
Are the benchmark ranges Duromex results?
No. They are industry reference ranges used for modelling. Duromex has no audited production results to publish yet — see case studies.
Ask for the pack.
ROI model, readiness guide, evaluation checklist and security pack — sent within one business day.
Duromex is pre-launch. Figures shown are design-partner targets and modelled economics, not audited results. Ask us for the methodology.