Starbucks and the AI inventory rollback: Computer vision failure and IT governance in physical retail

This teaching case examines the nine-month deployment and eventual discontinuation of Automated Counting, an AI-powered computer vision inventory management system developed in collaboration with Redmond, Washington-based firm NomadGo. One of the most prominent enterprise AI rollbacks in North American retail that year, the system was retired in May 2026 after being deployed across more than 11,000 company operated stores in North America starting in September 2025 due to ongoing accuracy issues, a doubled barista workload, and growing frontline frustration. As news of the retirement spreads, the case puts students in the shoes of Starbucks Chairman and CEO Brian Niccol on May 22, 2026. In order to preserve the trustworthiness of his “Back to Starbucks” recovery plan, Niccol must concurrently handle the outstanding inventory issue, revamp the company’s AI governance system, restore frontline trust, and inform analysts and institutional investors of the retreat. The issue has a secondary thread: Green Dot Assist, a generative AI chatbot that helps with recipes and equipment problems, was successfully launched while Automated Counting failed. This comparison offers students with a rich comparative perspective for investigating why AI systems succeed or fail in physical operations contexts, and what organizational variables impact those outcomes.