Starbucks Drops AI Inventory Tool in North America

• The AI tool miscounted inventory items, confusing similar milk types and failing to recognize products like peppermint syrup bottles on shelves
• Starbucks will return to manual counting methods for beverages and milk, the same approach used for other inventory categories
• The company is shifting focus to more frequent daily store replenishments and broader supply chain improvements rather than AI automation

Starbucks Discontinues AI Inventory Tool After Nine-Month Trial

Starbucks has terminated its AI-powered automated inventory counting program across North American stores, nine months after deployment, according to an internal newsletter reviewed by Reuters. The decision was announced Monday, May 21.

Key Details:

The AI tool was implemented as part of CEO Brian Niccol's turnaround strategy to address persistent product shortages that have negatively impacted sales. The system was designed to automate inventory counts for beverage components and milk products to improve visibility into stock levels at individual locations.

Performance Issues:

Reuters previously reported that the tool frequently miscounted and mislabeled items, including confusing similar milk types or failing to detect them entirely. Company video footage showed the system unable to recognize a peppermint syrup bottle while counting adjacent bottles on shelves.

Company Response:

Starbucks cited the need for "consistency and execution at scale" as the reason for retiring the program. The company stated it decided to "standardize how inventory is counted across coffeehouses." Going forward, beverage components and milk will be counted manually, using the same methods applied to other inventory categories.

Despite the tool's shortcomings, Starbucks had previously claimed the technology improved product availability—a key metric in Niccol's corporate turnaround efforts.

Future Plans:

The coffee chain indicated it will focus on supply chain improvements and transition toward more frequent daily replenishments to stores rather than relying on AI-driven inventory management.

This development highlights ongoing challenges in implementing AI solutions in retail operations and may signal caution for companies considering similar automation tools for inventory management.