Starbucks has discontinued its artificial intelligence powered inventory management system after just nine months of use, following repeated reports that the tool miscounted products, misidentified items and disrupted daily operations across its stores. The decision marks a notable setback for one of the world’s largest coffee chains as it continues to experiment with automation in retail environments.
The system, developed by technology provider NomadGo, was designed to track inventory of key beverage components such as milk, syrups and other supplies. It used automated scanning tools to help stores monitor stock levels and anticipate shortages. However, according to reporting from Reuters and statements referenced by Fortune, the system frequently produced inaccurate results, including what staff described as “hallucinated” inventories, where the tool failed to correctly detect items on shelves.
Starbucks confirmed the removal of the system, framing it as part of its ongoing process of testing new technologies and responding to operational feedback. A company spokesperson said, “We test ideas in our coffeehouses, listen closely to partner feedback, and make changes to deliver a better, more consistent experience.”
The decision follows months of internal frustration among store employees who said the tool created more work rather than reducing it. Some stores were required to reorganize storage layouts to align with the system’s scanning requirements, adding time and complexity to already busy workflows. Workers also reported that inaccurate readings led to supply chain inefficiencies, where stores either received too much stock or not enough of certain items.
Carl Addison, a Starbucks shift supervisor based in Shoreline, Washington, described how the system gradually became less reliable over time. He said, “It started off not particularly accurate and got less accurate over time.” His comments reflect broader concerns that the technology failed to improve in real world conditions despite initial expectations.
In some cases, the tool reportedly misidentified products entirely, creating mismatches between digital records and physical inventory. That meant store managers could not rely on its outputs when placing orders, undermining one of the core goals of the system, which was to improve efficiency and reduce waste.
The backlash from employees was also visible in internal feedback shared with the company. One barista response captured the frustration and relief felt by some staff after the system was withdrawn, stating, “Thanks for discontinuing Automatic Counting! The thought behind it was great, but the execution was proving difficult.”
Starbucks has been actively integrating artificial intelligence tools into its operations as part of a wider restructuring strategy under its leadership team aimed at improving customer experience and operational efficiency. These tools include systems that assist with recipe guidance, machine troubleshooting and order sequencing to reduce wait times. The company has argued that such innovations are necessary to modernize operations and address rising cost pressures in the retail sector.
Despite these ambitions, the inventory system’s failure highlights a broader challenge facing retailers adopting AI at scale. Many companies are finding that while AI can enhance certain processes, it often struggles in environments that require high precision and rapid human interaction. Retail settings, particularly food service operations, depend heavily on accuracy and speed, leaving little room for error.
Industry analysts have also noted that the rapid deployment of AI tools is sometimes driven by competitive pressure rather than proven operational value. As retailers race to adopt new technologies, the gap between expectations and real world performance is becoming increasingly visible. In Starbucks’ case, the tool was expected to streamline inventory tracking but instead introduced additional friction into store operations.
The company’s broader turnaround strategy, which includes store redesigns, menu simplification and digital tools, has shown mixed but improving results. Starbucks recently reported stronger quarterly sales growth in its US market, suggesting that other elements of its restructuring plan may be gaining traction even as certain technologies are rolled back.
However, the removal of the AI inventory system has sparked wider discussion across the retail and technology sectors about the limits of automation. Some experts argue that companies are still learning how to properly integrate AI into complex physical environments. Others believe that many deployments are premature and lack sufficient testing before being rolled out at scale.
A retail operations expert cited in Fortune’s reporting summarized the broader concern, stating, “Right now, there is more hype than actual benefit.”
The situation also underscores the importance of iterative development in AI systems, where companies refine tools over time based on real world feedback rather than expecting immediate transformation. Comparisons have been drawn with other retailers that spent years refining inventory systems before achieving meaningful gains.
For Starbucks, the experience serves as a reminder that even large scale investments in artificial intelligence must be carefully aligned with frontline realities. While the company continues to pursue automation across various aspects of its operations, the withdrawal of the inventory tool suggests a more cautious approach may be necessary when deploying technologies that directly affect store level workflows.
As AI adoption accelerates across global industries, the Starbucks case is likely to be studied as part of a growing list of examples where early enthusiasm for automation met practical limitations. It reinforces a central tension in modern retail innovation, balancing technological ambition with operational reliability.
