Walmart said in late July that it needed to rein in its sprawl of AI agents. After building dozens of separate tools, often accessed through different systems, the retailer admitted that the experience was getting confusing. Suresh Kumar, Walmart’s chief technology officer, described the new plan as consolidating everything into four “super agents”: one for customers, one for associates, one for sellers and suppliers, and one for developers. The logic was to stop requiring users to hop between narrow agents and instead offer single entry points that sit on top of the many task-specific agents running behind the scenes.
Just over a month later, Walmart has not slowed its pace. It has in fact introduced more agents, but in a more structured form. At its Converge 2025 technology event in Bangalore and through subsequent announcements, the company unveiled WIBEY, a developer-facing super agent, and reinforced that it already has customer, associate, and supplier super agents underway. CEO Doug McMillon told investors on the second-quarter earnings call that these agents will bring scheduling, sales data, supplier onboarding, and customer personalization into unified platforms.
Counting from July’s simplification plan to now, Walmart has committed to at least five super agents: Sparky for customers, Marty for suppliers, an Associate Agent for employees, WIBEY for developers, and another developer agent referenced by McMillon. Beneath these sit roughly 200 task-specific agents distributed across Walmart’s systems. The company’s Element machine learning platform has been upgraded to orchestrate these agents, offering stateful memory, reasoning capabilities, and cross-agent communication protocols. Walmart now treats agents the way it once treated models: catalogued, monitored, and reused across projects.
