Starbucks Retires NomadGo Inventory AI Across 11,000 Stores: Workers Had to Recount Every Scan

▤ Summary

Starbucks retired NomadGo's Automated Counting AI across more than 11,000 North American stores nine months after deployment. The LiDAR and tablet camera system, billed as 8x faster with 99% accuracy, frequently miscounted and mislabelled items, forcing employees to verify and re-enter counts and effectively doubling inventory work. Starbucks removed promotional materials, reverted to manual counting, and framed the change as a standardization decision, while NomadGo says it is learning from feedback. The case highlights the distribution-shift risks of physical-world computer vision and underscores that such tools require near-perfect accuracy to produce real labor savings for enterprise buyers.

▦ Linked Companies