After the Failure Following its AI Inventory Tool Deployment, Starbucks is Now Trying to Create its Own Internal AI Software – CoffeeTalk

Starbucks’ history with AI implementation took a notable turn in May 2026, following a failed deployment of NomadGo’s inventory AI tool across 11,000 stores. Despite its promise of 99% accuracy and speed, the system was plagued by operational failures, confusing milk varieties and missing products, ultimately forcing baristas to revert to manual inventory counting. CEO Brian Niccol’s initial confidence in this technology as part of a strategic turnaround plan proved misplaced.

Now, Starbucks is undertaking a new AI initiative focused on building its own internal software through AI-assisted coding, aiming to replace existing systems from Microsoft and IBM, which cost the company approximately $400 million annually. This new approach is framed by CTO Anand Varadarajan as a cost-cutting strategy, projected to realize around $30 million in savings in the upcoming fiscal year, with expectations for initial applications to roll out by late 2027.

The distinction between the failed NomadGo tool and current efforts is significant. The prior system’s failure stemmed from challenges in computer vision technology that were inadequately addressed. In contrast, the latest initiative seeks to create tailored enterprise software more efficiently, aided by advancements in generative AI coding tools, previously prohibitive due to associated costs and complexity.

Moreover, the broader context is crucial as Starbucks continues a $2 billion turnaround under Niccol’s leadership, reflecting a commitment not to vanity projects but rather strategic financial restructuring amidst improving business performance. The company’s Q2 2026 earnings indicated substantial growth, with consolidated net revenues rising by 9% and comparable store sales up 6.2%.

However, the execution of this new AI project must be approached with caution. Replacing critical operational systems invites risks, especially given Starbucks’ history of encountering challenges with technology deployment across its vast network of over 41,000 stores. Final approval of the internal software’s full deployment depends on testing results, a critical phase that may determine the project’s success or failure as it appears the market currently appreciates Niccol’s overall turnaround efforts based on stock performance, which increased by 27.92% year-to-date.

Read More @ The Street