Robot cuts clothing-folding time in half, XYZ says

XYZ has demonstrated the potential of physical AI technology for industrial applications.

The company recently used a physical AI-powered robot to cut clothing-folding time from 50 seconds to 25 seconds — a 50 percent reduction — while achieving a success rate of 96 percent. The achievement is seen as an example of improving not only task accuracy but also the speed and efficiency required for real-world service deployment.

XYZ unveiled the achievement at Nvidia AI Day Seoul 2026, held Tuesday and Wednesday at COEX in Seoul. The event brought together startups to share industry-specific AI use cases and the latest technology trends, while also exploring collaboration across the global technology ecosystem.

"For a robot to make autonomous judgments and carry out tasks, what matters is not just developing an AI model but building an integrated development framework — from collecting field data to training in a virtual environment and then executing on the robot," XYZ CEO Hwang Sung-jae said. "Based on this approach, we are advancing our physical AI development framework by leveraging Nvidia's AI computing platform and robot simulation technology."

Hwang added that the company uses data gathered from 50 sites operated by its subsidiaries to run verification and training in a virtual environment, then feeds the results back into the robots' task execution. XYZ plans to build this intelligence flywheel as quickly as possible to develop and commercialize robot intelligence capable of handling a wide range of working conditions.

"We organically connect customer interaction, task planning and physical manipulation through our proprietary robot intelligence framework, BrainX," Hwang said. "We are working to apply this to actual robot systems, including our semi-humanoid robot DEUX."