Changing what visual inspection AI can do
Turning the inspections that even AI could not handle into ones that run. Inspection Designer is an integrated AI inspection platform built on generative AI, with the industry-first Zero-Training AI that needs no training data and Ultra-Fine Good Product Learning built in.
- Free consultation and demo
- Customisation available
Inspection Designer
It starts with zero training.
An integrated generative-AI inspection platform with Zero-Training AI built in
You do not have to wait for defect images to pile up. Our own Zero-Training AI starts detecting with nothing learned in advance. Beside it sit Ultra-Fine Good Product Learning for defects as small as 2px, segmentation, object detection and AI OCR — eight AI types, all developed in house. Inspection flows are built without code, and the AI takes over the annotation, which runs five to ten times faster than annotating by hand.
Three things it does differently
- Training drags on with no end in sight
Zero-Training AIStart inspecting without collecting training dataOur own AI, covered by a granted patentNo gathering of defect images, no annotationA handful of images is enough to fine-tune it furtherSee details - Accuracy is too low to use on the line
Eight AI modulesMatch the AI to what you are inspectingEight AI types, rule-based checks and measurement, combined freelyDifferent conditions per region of the imageColour photometric stereo for surface reliefSee details - It goes in, and then nobody can run it
No-code by designKeep adjusting it yourself, on the floorBuild the inspection flow by connecting blocksOne-click annotationWatch results as they come inSee details
Inspections that were hard to automate, now running
FoodDiscoloured beans in the feed
The raw material varies widely in colour and size, so training is laborious — and a change of growing region can mean starting over.
Solved with Zero-Training AI
Metal & steelCoating defects on pre-painted steel
The products come in many bend shapes, and the combinations of shape and colour make inspection settings tedious to maintain.
Solved with Zero-Training AI
Printing & nameplatesTorn nameplates whose text keeps changing
The tear has to be found on a printed label, but changing lot information keeps being judged as a defect.
Solved with Ultra-Fine Good Product Learning AI
Chosen by manufacturers that lead their industries
From household-name finished-goods makers to the equipment builders who ship our software inside their machines. They span every sector, and they bring us the inspections that are hardest to automate.
- Aerospace
- Automotive
- Medical
- Electronics
- Food & beverage
- Cosmetics
- Textiles
- Metal & steel
- Semiconductor
- Moulded resin
- Logistics & packaging
- FOOMA JAPAN 2026
Top placing, startup zone - TOKYO DIGICONX The Pitch
Grand prix - Zero-shot anomaly detection AI
featured in ASCII - Tokyo AI × Data IP Acquisition
Support Project, selected
What customers tell us
- Zero-Training AI took our AI set-up from three months to ten minutes.
- Other vendors had turned us down over our product mix. OUEN had Zero-Training AI, so it felt worth one more try.
- I was surprised that defects this fine could be found from good products alone.
We build visual inspection AI ourselves, as an AI vendor
Since the early days of deep learning we have supplied manufacturers with algorithms we wrote ourselves. Selling is not the finish line for us; running on the line is. Alongside visual inspection we take on contract development with large language and image generation models, and we run AI training for companies.
