Oxford University spinout Aistetic developed an AI body-scanning tool intended to make online clothing sizes more accurate. A shopper could record a short turn using a phone, tablet or computer, receive measurements and obtain retailer-specific sizing in less than three minutes.
The product addressed an expensive mismatch between digital shopping and physical fit. Poor sizing leads to returns, lost margin and additional transport, while consumers waste time ordering several options to discover which garment works.
A Short Mobile Scan Replaced Manual Measuring
Users stood back from a camera and turned for around ten seconds. Aistetic said its system could derive body measurements with 98 percent accuracy and then match them to sizes used by the retailer hosting the tool.
Convenience was central. Many customers do not own a measuring tape or know how to take consistent garment measurements. A guided recording can standardise the input while making the experience available on common devices.
Accuracy claims still need testing across body shapes, clothing, lighting, camera quality and mobility. A system that works well in ideal demonstrations must remain usable in ordinary homes and should recognise when the scan is insufficient rather than providing false precision.
Low-Code Integration Reduced Retailer Friction
Aistetic offered a short JavaScript snippet for websites using systems including WordPress and Shopify. Shoppers could select a garment, scan themselves and receive sizing within the existing store journey.
This approach avoided asking retailers to rebuild their commerce platform. It also allowed sizing to reflect each seller’s own chart, an important feature because a nominal size can vary significantly between brands and products.
The company planned a fully no-code Shopify version distributed through the platform’s app marketplace. Easier installation could expand access among smaller retailers that lack dedicated development teams.
Simple installation should still include clear configuration checks. Retailers need to map every garment to the correct size chart, test the mobile experience and assign ownership for updates. Low-code describes technical effort, not the absence of operational responsibility.
Better Fit Could Reduce Returns and Emissions
Aistetic said the tool could reduce return rates by up to 30 percent. Even a smaller improvement could matter for online retailers, where reverse logistics, inspection and resale reduce the value of each transaction.
Returns also have an environmental cost. The source cited an estimate that fashion contributes 10 percent of global carbon emissions and said returning one garment to a store could create as much as 4.2 kilograms of carbon.
Accurate sizing will not solve fashion’s full environmental impact, but it targets a specific avoidable source of transport and handling. Retailers should measure actual return reductions by product type and ensure that sustainability claims reflect observed outcomes.
Body Data Requires Consumer Control
The founders described an ambition to make next-generation 3D body modelling available to anyone with a mobile device and to empower people with their own measurements. That promise depends on treating scans as sensitive personal data.
Customers need to know whether video is uploaded, how long measurements are retained, who can access them and whether the information is reused for model development. The safest design collects only what is needed for sizing and gives people simple deletion controls.
Retailers should not turn measurement data into an unrelated profiling tool. A consumer seeking better fit has not automatically agreed to broader inferences about health, identity or marketing preferences.
The company was founded in 2019 by Duncan McKay and Oxford computer-vision professor Phil Torr. It received Innovate UK grants and a Future Fashion Factory grant with the University of Leeds, totalling about £1.2 million.
That research and commercial background helped connect computer vision with a concrete retail problem. The product’s value did not depend on presenting AI as futuristic; it depended on saving shoppers and merchants from an everyday frustration.
Adoption should be evaluated through completion rate as well as measurement accuracy. If scanning feels awkward, intrusive or difficult, customers may abandon the process even when the resulting size is reliable.
Inclusive design can provide alternatives for people unable or unwilling to perform a full turn. Clear instructions, feedback on camera placement and a manual-measurement route prevent the tool from becoming a new barrier.
Aistetic illustrated how applied AI can create value through a narrow, measurable task. The strongest case was not that computer vision could model the body, but that it could make clothing selection more dependable while reducing unnecessary returns.
If the system combines accuracy, easy integration and genuine data control, mobile body scanning could improve the economics and experience of online fashion. The consumer should remain the owner of the intimate information that makes that convenience possible.
Retailer-specific recommendations also need maintenance. Size charts change, garment cuts vary and two products carrying the same label may fit differently. Integrations should connect measurements to current product data and monitor where suggested sizes still result in returns.
Feedback can improve the service if it is collected carefully. A shopper might confirm whether a recommended item felt tight, loose or correct, but that information should be voluntary and separated from unrelated advertising profiles. Product improvement does not require unlimited retention.
The experience should communicate probability rather than certainty. Bodies and clothing preferences cannot be reduced to one perfect number, and some customers deliberately choose a looser or closer fit. Presenting a recommended size with a short explanation lets the shopper retain judgement.
For retailers, the most convincing evidence would be controlled comparison across categories and customer groups. Return reduction, conversion, satisfaction and scan completion should be measured together. A tool that lowers returns by discouraging purchases would not create the same value as one that improves confident fit.
