Automotive supplier Yazaki North America has agreed to deploy DeepHow’s Time and Motion AI platform across its production lines, starting with operations in Mexico, in an engagement facilitated through Nvidia. The system automates cycle-time capture and workflow analysis, replacing manual time-and-motion studies that previously took engineers weeks to complete.

The platform integrates with Nvidia’s Metropolis Blueprint for video search and summarisation and Nvidia RTX Pro Servers. Powered by vision-language models, the system captures cycle times and classifies workflow segments in real time, surfacing waste reduction opportunities and supporting process improvement across production cycles.

Yazaki North America expects to cut line-analysis time from weeks to days, with projected annual savings running into the millions of dollars from Mexican operations alone. At global scale, the firm anticipates savings in the tens of millions annually.

The deployment enables on-demand analysis of any production line rather than scheduled studies a few times per year, allowing earlier identification of bottlenecks and faster work rebalancing. DeepHow is also developing automatic report generation and factory flow intelligence built on Nvidia Cosmos.

The data layer the system creates represents a longer-term opportunity. Factories typically generate extensive machine data but little structured information about the sequencing, variations and decision points in manual assembly workflows that influence quality and throughput. By capturing those dynamics in structured form, Yazaki North America begins building a continuous record of its production methods, available for ongoing analysis and cross-site improvement.

Source: DeepHow