Forming & stamping
Adaptive press-parameter control with seam and weld quality scoring, and dent prevention rather than dent detection.
- Metric: scrap rate and rework at forming
Autonomy thatknows what itis building.
A refrigerator cabinet, a washer drum and an enamelled oven cavity fail in completely different ways. Duromex is vertical by design — the models, the recipes and the standards are appliance-specific, not generic vision.
Five categories, five different failure modes
Pick the product family you build. The wedge workflow, the models and the ROI metric all change with it.
Refrigeration
Cabinets are formed, foamed and charged — then judged by a leak test and an energy rating. Duromex links the foam shot and the seam to the number the tester prints.
- Foam density, distribution and void prediction per cabinet
- Refrigerant leak signature detection before charge-and-seal
- Energy-performance forecast from process conditions, not just test results
- Door seal and gasket compression verification
Laundry
Drums, cabinets and wiring harnesses at high volume. Balance, seal and electrical integrity decide whether a machine survives its warranty period.
- Drum and cabinet forming control with seam-quality scoring
- Water-leak detection at fill, drain and seal stations
- Harness routing, connector seating and polarity verification
- Vibration and spin functional-test triage
Dishwashing
Tub forming, insulation, pump and heater assembly, then a wet functional test that is expensive to repeat.
- Tub seam and weld integrity sensing
- Water-leak and overflow-path detection
- Pump, heater and sensor assembly verification
- Wet-test cycle-time and repeat-test reduction
Cooking
Enamel is the product. Coverage, film build, cure and chipping decide whether a range ships as first quality or a downgrade.
- Enamel coverage, film build and cure control
- Chip, pinhole and orange-peel detection
- Gas valve, igniter and element assembly verification
- IEC 60335 safety-test evidence capture
Air conditioning
Brazed joints, refrigerant charge and performance test under tightening F-gas rules — a leak here is a compliance event, not just a rework.
- Brazed-joint and coil leak detection
- Charge accuracy and refrigerant loss tracking
- Performance and capacity test triage
- F-gas and energy-regulation evidence trail
Or start at the stage that is bleeding
Most plants know exactly where the money leaves. Duromex lands on that stage first — forming, enamel, foaming, assembly, leak and functional test, or packing — then expands across the loop as the ROI proves out.
Form & stampWeld & seamEnamel & paintFoam & insulateWire & assembleLeak & function testRate & pack
What each stage gets on day one
Every stage ships with its own perception models, control interface and success metric.
Enamel & paint
Coverage, film build and cure control with run, sag, orange-peel, chip and pinhole classification.
- Metric: first-quality rate and downgrade rate
Foaming & insulation
Shot weight, mix ratio, jig and cure control with void and density-gradient prediction per cabinet.
- Metric: energy-rating conformance and foam rework
Assembly & wiring
Placement, torque and connector verification for components, harnesses and electronics on every unit.
- Metric: field-failure and recall-risk reduction
Leak & functional test
Refrigerant and water leak detection fused with functional, performance and energy testing, plus automated triage.
- Metric: functional-test failure and warranty rate
Robotic handling & packing
Cobot task allocation, material-handling flow, pack and label verification with robot-zone safety supervision.
- Metric: labour dependency and pack error rate
Three people have to say yes
The Plant or Operations Director is the economic buyer — they own first-pass yield, throughput, warranty and cost. The Manufacturing or Process Engineer and the Quality or Test Engineer are the champions who live with the outcome. IT/OT, quality and security approve.
Duromex is built so each of them gets the thing they actually need, rather than a single demo aimed at whoever is in the room.
What each one needs
- Director: hard ROI in a paid pilot, tied to the pricing metric
- Process engineer: autonomous forming, foaming and assembly relief
- Quality engineer: IEC 60335 traceability and test acceptance evidence
- IT/OT: SSO, RBAC, isolation and an on-prem option
- Security: immutable audit trail and strict recipe IP protection
The six pains, and what replaces them
These are the phrases that come up in every discovery call — and the measurable outcome Duromex is priced against.
Test failures & end-of-line rework
Failures predicted upstream and attributed to a station, so the fix happens where the cost is still small.
Refrigerant & water leaks
Micro-leak signatures caught below tester thresholds, before charge, seal and ship.
Surface, weld & enamel defects
Adaptive control that prevents the defect instead of a camera that counts it.
Foam voids & density errors
Thermal performance forecast per cabinet, so the energy rating stops being a surprise.
Wiring & component errors
Verified placement, torque and connector seating on every unit, with evidence retained.
Labour turnover & high mix
The craft encoded in models, and changeovers rehearsed in the twin before the line meets them.
Which agent owns which problem
Land on one. Expand across the rest as the loop closes.
Who Duromex is — and is not — built for
We would rather disqualify early than run a pilot that cannot succeed.
| Strong fit | Poor fit | |
|---|---|---|
| Company | Appliance and white-goods OEMs and tier-1 suppliers, $100M–$40B revenue | Importers, rebranders and assemblers with negligible in-house manufacturing |
| Volume | Enough production volume for clear ROI on one line | Low-volume or job-shop production |
| Systems | Forming, paint, foaming, leak-test, assembly and MES with API/OPC access | Sites unwilling to integrate under security controls |
| Mandate | Executive mandate to lift first-pass yield or cut warranty, scrap and rework | Exploratory interest with no owner and no metric |
| Expectation | A system of action that performs the work | A generic horizontal analytics dashboard |
Land narrow. Expand relentlessly.
$12k
per line, per month — the wedge
$80k
per factory, per month — the loop
$7M
top of the enterprise ACV land range
135%
target net revenue retention
Expansion is the primary growth engine: more lines, more modules, more sites, and more autonomy at each one.
From the three people who have to say yes
“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.”
What the first year usually looks like
The sequence is deliberately boring. Boring is what gets a vendor write access to a press.
Weeks 1–4
Instrument and baseline
Connectors go in on one line. Shadow mode runs. We publish the measured baseline for your chosen metric.
Months 2–3
Advisory mode
The agent recommends and predicts; your engineer approves. Accuracy and ROI are proven against the baseline.
Months 4–6
Graduated autonomy
Low-risk control and automatic defect and leak triage go live. Exceptions escalate. The twin gates every change.
Months 7–12
Second module, then the factory
An adjacent module in the same plant, then the full loop — and the conversation moves from one line to the site.
Choosing where to start
We have several bad lines. Which one should we pick?
The one with the largest, most measurable loss and the best data access. A pilot succeeds or fails on whether the metric is unambiguous, not on how interesting the AI problem is.
Do you support high-mix plants with frequent model changes?
That is one of the strongest cases for the twin. Changeovers are sequenced and rehearsed in simulation, and recipes carry over between product variants rather than being re-tuned by hand.
We build for several brands on one line. Is that a problem?
No. Product, BOM and recipe grounding is per product, and the orchestrator plans against the unit actually on the line.
Start where it hurts. Expand where it pays.
One line, one workflow, one metric. Then the loop.
Duromex is pre-launch. Figures shown are design-partner targets and modelled economics, not audited results. Ask us for the methodology.