CATCHES Launches AI Sizing Technology for Fashion Ecommerce

CATCHES is solving one of the biggest problems in fashion ecommerce – customers not knowing whether a garment will fit them.

CATCHES, an AI company developing application and infrastructure layers for digital fashion experiences, has launched its generative AI Sizing technology, an AI-based online consumer experience built on NVIDIA’s CUDA platform, designed to provide sizing, fit, and drape information for fashion ecommerce.

RealFit, announced at NVIDIA GTC, is a capability that allows shoppers to view a representation of how a garment may fit, drape, and move on their body before purchasing. 

Shoppers enter their measurements and a photograph to generate a digital twin, then view garments in the collection on that model, toggling between sizes to understand how each piece is expected to fit. The first public use of CATCHES’ RealFit went live on the AMIRI website. CATCHES is currently working with a number of other brands, scheduled to launch in the next few months.

ALSO READ: The State of Customer Experience: What 2025 Taught Us

“Accelerated computing is unlocking new possibilities for digital fashion experiences. CATCHES has introduced an AI–based virtual try-on with Physics-Based simulation for fabric movements and sizing, built on NVIDIA Omniverse and an accelerated computing platform. AMIRI is pioneering this innovation, and we’re collaborating with CATCHES as an Independent Software Vendor to bring these capabilities to more fashion brands,” said Azita Martin, VP and GM of AI for Retail and CPG, NVIDIA.

“CATCHES is solving one of the biggest problems in fashion ecommerce – customers not knowing whether a garment will fit them. Fit uncertainty reduces online conversions and leads to over 50% return rates in some categories. CATCHES RealFit technology fixes this by providing customers a personalised photorealistic experience that understands real materials, cuts, and true sizing & fit. With the support of NVIDIA, we are providing brands a fully customised offering and a customer experience for the AI era,” said Ed Voyce, Founder and CEO, CATCHES.

ALSO READ: UK Consumers Call on AI to Save “Broken” Customer Service

CATCHES has raised $10 million from the world’s leading luxury voices, including Antoine Arnault, Natalia Vodianova Arnault, Roy Chung, founder of Apollo.io, Dillon Erb, founder of Paperspace, Gary Sheinbaum, former CEO for Tommy Hilfiger, and Sarah Willersdorf, former Head of Luxury for BCG.

“In fashion, the relationship between a garment and the person wearing it is essential. Any technology that helps bring that understanding into the online environment is an important step forward,” said Antoine Arnault, Director of Image & Environment at LVMH.

CATCHES spent two years building a proprietary GPU-accelerated simulation framework, modelled and integrated on physical fabrics, before applying AI to scale it. The goal is to closely represent the fabric’s weight, structure, drape, and movement. 

ALSO READ: Phonexa Increases Consumer Conversions Through SMS

For shoppers, RealFit is intended to approximate the experience of a fitting room from a remote location. For brands, it offers an opportunity to strengthen the connection between customer and product, potentially increase purchase conversions, and reduce fit-related returns, which are a major driver of reverse logistics costs across fashion ecommerce.

The virtual experience is built on NVIDIA CUDA and Omniverse libraries, leverages the Newton Physics Engine, and combines custom-trained diffusion models with fine-tuned VLMs and LLMs, including AI models based on NVIDIA Nemotron and NVIDIA Cosmos world foundation models, to support physically grounded generative experiences. 

CATCHES’ bare-metal infrastructure is distributed across NVIDIA RTX PRO™ 6000 Blackwell Server Edition, NVIDIA Hopper, and NVIDIA Blackwell and Blackwell Ultra GPUs, with at-the-edge inference provided by Nscale, a vertically integrated AI hyperscaler, and through bare-metal partners Nebius, Boston Limited, and Scan Computers.

The system pairs AI with NVIDIA CUDA and Omniverse-accelerated physics simulations to generate digital twins designed for millimetre-level fit detail. By building the solution on an OpenUSD foundation, data can be exchanged between the digital twin model, Blender, and Unreal using NVIDIA RTX Ray. 

ALSO READ: PropellerAds Reveals Social Traffic Targeting for Increased Conversion Rates

- Advertisement -spot_img

Featured Articles