How Walmart uses multi-agent AI to predict grocery store inventory

Aditya Mulik, a senior software engineer at Walmart Global Tech, presented a case study on using multi-agent LLM workflows to solve the massive retail challenge of inventory accuracy, particularly for perishable goods like rotisserie chicken and fresh beef across Walmart's 4,700+ U.S. stores. The system aims to predict exactly what inventory each department needs for upcoming days at enormous scale, automatically balancing the need to prevent empty shelves against the waste caused by overstocking.

The system addresses a critical operational problem where inaccurate predictions directly lead to lost sales from stockouts or financial losses from spoiled food that must be discarded. By applying cost-efficient AI workflows with built-in resilience, Walmart is attempting to automate what has traditionally required extensive manual forecasting by store managers, representing a significant real-world application of multi-agent AI to a complex logistics challenge with immediate business impact.