Walmart Lets AI Agents Shop as Anthropic Launches Claude Duet

Walmart said its Agent Gateway is a new MCP server that connects approved agents to its site through the Model Context Protocol, an open standard originally created by Anthropic. The setup lets an agent search Walmart’s online inventory and make purchases for its user. Walmart said the gateway will also help customers buy their usual brands or items without sharing their broader shopping history or personal information directly with the agent. The company said the technology will support agentic shopping on both Walmart and Sam’s Club.

Agent Profiles, rolling out next month, will let customers authorize each AI agent through a dedicated profile in their Walmart account. Purchases made by an agent will appear separately from the customer’s own, and customers can require approval for every transaction or set a spending limit for an agent. Walmart compared the controls to adding an authorized user to a credit card. Agent Profiles can be created the first time a customer uses an agent through the gateway, and customers can delete a profile at any time. Walmart said any customer with a Walmart account can use the feature; a Walmart+ membership is not required.

Walmart also published terms for AI agents. The retailer said agents must not take actions beyond what the shopper intends, cannot change a customer’s payment method or access payment information, and can only access information needed for a given task. Agents that do not comply will be blocked. Walmart said its returns process applies, including when an agent does not act as intended. The move contrasts with Amazon, which recently began blocking AI agents from making purchases; The Verge reported that Amazon blocked Perplexity’s shopping agent and Meta’s Muse. Walmart has already announced a partnership with Meta’s Muse and offers integrations with ChatGPT and Gemini, according to TechCrunch. Walmart President and CEO John Furner, speaking at the Masters of Scale Summit in San Francisco, said the company was preparing for a world where AI agents become mainstream.

Anthropic said Claude Duet is designed for professionals and small business owners and works by messaging users the way a person would, operating across connected apps, and asking before it sends a message, makes a payment, signs a document, or deletes anything. It runs on Claude Opus 5.5 and is available through standalone iOS and Android apps and on the web for users who receive an invitation. Anthropic said Duet runs on its own cloud-based virtual machine, allowing it to keep working after a user closes the app, and uses the same memories, skills, and connectors as regular Claude. Each morning it sends a brief on items that need attention; larger jobs can return as files such as research briefs, financial models, slide decks, or dashboards.

Duet is available as a research preview to Pro and Max subscribers in the US through a waitlist, Anthropic said. The feature is included at no additional cost but counts toward existing plan usage limits, including background work, which the company said could cause users to reach limits sooner than with chat alone. Engadget reported that Duet will be available to subscribers of Anthropic’s $20-per-month Pro plan when it reaches general availability, while OpenAI’s Dots are currently limited to its Pro tier, which starts at $100 per month. Anthropic said Duet has a new classifier and system prompt intended to prevent it from sharing the strictest tier of information, such as credit card numbers, and that it will not directly see passwords or sign-in tokens. The company described Duet as an early version that may misread a thread or miscalibrate what is urgent and said it is not as strong at personal errands such as booking tickets.