Fashion’s Other AI Revolution

6 Nov 2025

A typical Walmart store manager oversees hundreds of employees and frequently handles promotions, hires, and terminations. To simplify this, Walmart built an AI agent that executes HR transactions autonomously. Managers simply issue requests like “promote associate,” and the agent completes them without human intervention, according to Ben Peterson, VP for people product and design.

Walmart is an early pioneer in using AI agents in retail, leveraging large language models to perform tasks autonomously. While agents have drawn attention in online shopping, their immediate impact is in back-office roles. CEO Doug McMillon noted that AI will transform every job, even as Walmart keeps its 2.1 million-employee headcount stable.

Other companies, including Amazon and Target, are also exploring AI to improve efficiency, often through automating repetitive tasks. Walmart, for instance, has built a merchandising agent that produces performance summaries, tracks sales trends, and diagnoses under- or overperforming products.

AI adoption is still limited. Research from Carnegie Mellon found that even the best open-source agents can complete only about 30% of realistic company tasks autonomously. Complex, tedious tasks remain challenging, but technology is improving rapidly.

Enterprise platforms like SAP, Salesforce, and Workday now offer agent solutions for businesses. These agents can automatically generate purchase orders, plan marketing campaigns, or adjust order routing, often with minimal coding required. Some claimed “agents” are actually LLM-powered chatbots supplemented with retrieval-augmented generation (RAG), which is a more limited form of automation.

True agents, according to BCG’s Matt Kropp, need three capabilities: reasoning and planning via LLMs, the ability to use tools to execute actions, and memory beyond the immediate conversation. When fully realized, agents could navigate legacy systems and complete complex tasks without APIs, although current agents still struggle with complicated interfaces and long workflows.

Currently, the most advanced use of agents is in software engineering, but they are increasingly applied in customer support, marketing, and sales. Innovative use cases include “synthetic consumer research,” where multiple agents simulate customer behavior to provide actionable insights faster and cheaper than traditional surveys.

Walmart envisions a future where multiple agents collaborate to handle complex tasks seamlessly. Peterson predicts that current limitations will likely disappear within 6–12 months as the technology and market evolve rapidly.