Walmart opens its doors to AI shopping agents with guardrails and spending caps

Walmart is moving forward with a plan to permit AI agents to conduct shopping on its platform, a strategy unveiled by Chief Executive Officer John Furner at the Masters of Scale Summit in San Francisco. The initiative stands in contrast to Amazon's approach, which has blocked comparable agents from accessing customer accounts. Walmart's framework establishes strict boundaries, preventing AI agents from accessing payment card information and restricting their actions to what customers explicitly authorize.

Separate profiles with individual limits

Currently, when an AI agent makes purchases through Walmart.com or Sam's Club, those transactions appear under the customer's personal account. Agent Profiles, launching within the coming month, will change this by creating distinct profiles for each agent within a customer's account. Shoppers retain the ability to either approve each transaction individually or establish predetermined spending thresholds—such as $50, $100, or $200—within which agents can operate autonomously. Customers maintain the option to remove any agent profile at any time.

Think about it like adding someone as a cardholder on your credit card

Walmart

The system initializes when a customer first engages with an agent through Walmart's gateway, after which the customer authenticates by logging into their Walmart account. Notably, a Walmart+ subscription is not required to use this functionality.

A specialized gateway for machine agents

At present, AI agents navigate Walmart's website using conventional web browsers, executing the same steps designed for human shoppers—a process Walmart characterizes as inefficient and prone to errors. The newly introduced Walmart Agent Gateway bypasses this approach by utilizing the Model Context Protocol (MCP), an open technical standard developed by Anthropic to facilitate connections between AI agents and external systems and information sources. The gateway is launching as a beta offering for companies and software developers.

One advantage of this architecture is that agents can identify a customer's frequently purchased items—for instance, a preferred brand of milk—without gaining access to the customer's complete purchase history, according to Walmart.

Strict operational guidelines

Walmart has established a comprehensive set of rules governing agent behavior on its platform. By default, customers must authorize each purchase made by an agent, though Walmart reserves the right to require additional approvals even when transactions fall within an established spending limit. Agents are explicitly prohibited from viewing payment card details or modifying payment methods. Agents must restrict their access to only the information necessary for completing a specific task and cannot exceed the boundaries of what the customer intends.

The standard Walmart returns policy applies to all transactions, including those executed by agents that fail to perform as expected. Agents that violate these terms face removal from the platform.

Walmart has committed to maintaining consistent pricing regardless of whether a purchase is made by a human or an AI agent.

We price products, not people, so the prices that an AI agent sees are the same prices that a human would see on Walmart.com, with applicable offers and fees shown before purchase

Walmart

In September, Furner explicitly rejected the possibility of implementing pricing that varies based on individual customer characteristics. Lindsay Owens, director of the Groundwork Collaborative, has publicly criticized Walmart for engaging in what she characterizes as deceptive pricing tactics, according to reporting by Axios. Walmart has disputed these allegations, describing them as inaccurate and misleading.

A different strategy from Amazon

Walmart already permits purchases through several AI platforms, including Meta's Muse, OpenAI's ChatGPT, and Google's Gemini. Amazon has taken the opposite stance, prohibiting Meta's Muse from accessing its ecosystem in September. Amazon stated that it had not granted Muse authorization to connect to customer accounts, extract customer data, or handle financial transactions, as reported by Axios. Financial institutions including Schwab and JPMorgan have similarly blocked AI agents from managing customer funds.

Source: The Next Web