Starbucks scraps – Starbucks is scrapping its AI-powered inventory counting system across North America just nine months after it was deployed, after cameras and LIDAR-equipped tablets repeatedly produced miscounts and labeling errors—confusing similar milk types and even missin
A coffee shop routine that sounds almost mechanical—counting what’s on the shelf—turned out to be the hardest part to automate.
Starbucks is officially scrapping its AI-powered inventory counting system across North America just nine months after deployment. The system was built to use cameras and LIDAR-equipped tablets to scan beverage inventory and ingredient stock across stores. with the goal of automating stock counting and reducing in-store shortages.
But the tool reportedly struggled with frequent miscounts and labeling errors. Workers faced mistakes that confused similar milk types or resulted in products being missed entirely. Employees reportedly continued to battle inaccurate counts and unreliable product recognition even after the company previously said the system improved inventory visibility.
The shift is tied to CEO Brian Niccol’s “Back to Starbucks” turnaround strategy, which included improving product availability and operational efficiency.
In internal messages reviewed as part of the reporting, employees reportedly even celebrated the tool’s removal. Starbucks now says it will return to manual inventory counting while focusing on more standardized replenishment systems and daily restocking improvements.
The sequence is hard to miss: an AI system meant to handle structured. repetitive inventory work was deployed to real stores. where lighting. packaging similarities. and day-to-day chaos are unavoidable—and accuracy fell apart quickly enough to trigger a rollback within less than a year. When a system designed to “see” what’s inside a coffee shop keeps getting it wrong. the gap doesn’t just show up in dashboards. It lands in the middle of work that still has to be done.
For a company betting on AI in operations. this is a particularly pointed moment—one that underscores how quickly the real world can outpace a polished promise. Starbucks is choosing the boring solution again: people counting, tighter replenishment, and daily restocking improvements instead of automated scanning.
Starbucks AI inventory system AI inventory counting LIDAR tablets milk carton miscounts Back to Starbucks Brian Niccol retail automation in-store shortages