Walmart: 4,600 Stores, Challenging AI Returns to Manage

The distributor equips its in-store teams with AI tools and expects them to help shape these tools. With over 4,600 points of sale, maintaining employee engagement and sorting through divergent feedback becomes a central issue. Concrete cases show friction, while management acknowledges mistakes and is already citing internal successes.

Involving 4,600 stores requires balancing heterogeneous feedback

Managing employee feedback becomes a challenge in itself when the collection occurs within a chain that includes more than 4,600 stores across the United States. The needs of a location in Philadelphia do not necessarily align with those of a store in New Orleans, making divergent opinions on the same tool inevitable. A decision must then be made about which feedback to heed and which to disregard.
Walmart's Vice President for Associate Tools states that he wants to give each store the latitude to use the tools as they see fit, an intention that clashes with the difficulty of navigating contradictory feedback. Another hurdle is maintaining team engagement so that they continue to share their opinions over time, especially since obtaining responses to surveys is often challenging.

In-store tools that can burden the workload

Workers report that some tools do not account for the practical complexities of tasks like restocking, which may involve cleaning up a mess or removing expired products. In other cases, the tool ends up adding unexpected steps.
Spark, Walmart's delivery service, recently updated a feature to guide drivers in stores while collecting items. According to employees, real-time location tracking has caused them to lose time and sometimes led them to pick up products like ice or frozen goods first.

Management acknowledges mistakes and promotes "AI for all"

Walmart believes that mistakes are part of the adoption process and expects its teams to participate in shaping the tools they use. John Furner has previously advocated for an "AI for all" approach, deemed more effective than a centralized directive.
This direction is also based on successes. Leo Garcia, a logistics manager, developed an AI application that, according to him, has allowed drivers to return home earlier while reducing the number of empty trailers.

A large-scale participatory ambition, between promise and constraints

Walmart is deploying AI tools among its store teams with the ambition that employees contribute to their continuous improvement. However, some report often needing to correct errors and train the technology. As the largest private employer in the United States, the company highlights both the strengths and limitations of an approach involving end-users in development: it fosters a connection with the realities on the ground but also confronts the two mentioned obstacles, namely maintaining engagement and managing feedback.