The post names Shopify, Zooplus, Photoroom, Otto Group, CATCHES, and other companies.
Shopify said its post-trained Nemotron 3.5 Lightning GraphQL subagent for Sidekick cut serving cost by 90% and returned responses seven times faster on an internal benchmark. It also reported up to 4.5x faster performance and up to 12x lower inference cost across five additional uses. Zooplus reported 86% search accuracy for a production-ready prototype in its AWS environment. Photoroom said it processes about 40,000 listing images per day for Vestiaire Collective across 70 countries.
For physical operations, NVIDIA said Otto Group extended a simulated, human-supervised autonomous-warehouse project to its largest fulfillment center in Haldensleben, Germany. CATCHES reported a 55% conversion increase in its RealFit deployment with AMIRI. NVIDIA also said three retail Blueprints shown by Nebius were available in early access. The source does not provide independent validation of the reported benchmarks, commercial terms, or evidence that the results will transfer to other retailers.