Starbucks — the global coffeehouse chain headquartered in Seattle — has withdrawn its artificial intelligence system for automatic inventory counting after months of operational failures in North American stores. The decision comes after multiple incidents that affected the availability of basic products such as milk, syrups, and other key supplies for daily service.
The system, developed as part of the company's operational modernization strategy, aimed to reduce manual workload in establishments through computer vision and sensors.
However, in practice, the software generated recognition errors, confusion between similar products, and incomplete counts.
A system designed to streamline, but which ended up complicating work
The automated inventory program, deployed approximately nine months ago, used cameras and scanning technology to identify products on shelves. The idea was simple: reduce human errors and improve real-time logistical efficiency.
But according to internal reports and testimonies collected by media such as Reuters, the system frequently confused types of milk, omitted items, or recorded incorrect quantities. This led to discrepancies in product restocking and, in some cases, stockouts in stores.
Given the accumulation of problems, the company decided to return to manual counts, performed by in-store employees, as the primary method of inventory control.
Impact on the company's modernization strategy
The withdrawal of the system represents a setback within the internal transformation plan driven by CEO Brian Niccol, who has bet on automation, artificial intelligence, and logistical optimization as pillars to recover operational efficiency.
The "Back to Starbucks" strategy precisely aimed to improve in-store consistency, reduce waiting times, and avoid product stockouts. However, this episode highlights the difficulties of applying artificial intelligence in complex physical environments, where real working conditions are less predictable than in digital environments.