Starbucks has shut down an ambitious artificial intelligence inventory tool across its North American stores after just nine months in service, returning milk and beverage components to manual hand counts as of May 18, 2026. The company confirmed the move in a May 21 statement to Reuters, saying it would now “standardize how inventory is counted across coffeehouses” as it focuses on “consistency and execution at scale.”
The AI-based “Automated Counting” system had been rolled out to more than 11,000 company-owned locations across North America starting in September 2025, according to a launch release from BusinessWire, while a separate NomadGo video cited 11,300 stores. Built by Seattle start-up NomadGo, the system used tablet-mounted cameras, LiDAR, computer vision, three-dimensional spatial intelligence and augmented reality to scan shelves and count key ingredients.
When it was introduced, Starbucks’ chief technology officer Deb Hall Lefevre described the tool as a way to streamline “a critical but time-intensive task.” In the September 2025 launch announcement reported by BusinessWire, she said the company expected “faster, more accurate inventory counts” that would let staff “spend more time focusing on what matters: crafting high-quality beverages and connecting with customers.” That blog post has since been removed from Starbucks’ website, according to coverage by Engadget.
NomadGo promoted the deployment in the same BusinessWire release as “a strong validation” of its patented technology, claiming the system delivered “up to eight times faster results than manual methods with 99% accuracy.” NomadGo co-founder and CEO David Greschler said at the time that its platform “not only streamlines inventory management but also delivers frequent, highly accurate data that prevents out-of-stock situations and lowers cost of goods sold.”
In practice, reports gathered by The Globe and Mail, TNW and Engadget describe a less reliable picture. According to those outlets, the app frequently miscounted or mislabeled items, confusing similar milk types such as oat and dairy milk or missing them altogether. A promotional video shown at launch and cited by The Globe and Mail, Engadget and MarketScreener showed the system failing to register a bottle of peppermint syrup on the shelf while counting adjacent bottles.
An internal Starbucks newsletter dated May 18, obtained by The Globe and Mail, told staff: “Starting today, Automated Counting will be retired. Beverage components and milk will now be counted the same way you count other inventory categories in your coffeehouse.” Engadget, citing internal notes, reported one employee responding, “Thanks for discontinuing Automatic Counting! The thought behind it was great, but the execution was proving difficult.”
The inventory tool was part of CEO Brian Niccol’s “Back to Starbucks” turnaround plan, aimed at addressing chronic stock reliability issues, according to TNW. TNW, citing Reuters, reported that in 2024 fewer than a third of deliveries to Starbucks distribution centres arrived on time and in full, and that four Starbucks CEOs over five years have linked lost sales to difficulties keeping stores properly stocked.
Operationally, Starbucks enters this shift away from AI counting with mixed financial signals. Reuters-based reporting in The Globe and Mail and MarketScreener notes that the company’s North American operating margin has declined from 18% two years ago, before Niccol took over, to 9.9% as of May 2026. Over the same period, Starbucks recorded its strongest quarterly revenue growth in two and a half years in the first quarter of 2026, with global same-store sales rising 4%, and its stock was up 24% year-to-date by late May.
Looking ahead, coverage by The Globe and Mail and TNW states that Starbucks will continue to count milk and beverage components by hand, using the same approach as other inventory categories, while working toward more frequent, daily replenishments and broader supply chain improvements. In a statement to Reuters carried by The Globe and Mail, NomadGo said only that it is “continuously learning from customer and user feedback to improve our products.”
