- Section
- Ecommerce & Digital
- Published
- 4 min read
- By James Calloway
Ecommerce & Digital
Walmart Gives Shoppers Spending Limits and Controls Over AI Agent Purchases
Walmart will launch Agent Profiles within a month, letting shoppers cap AI agent spending at $50–$200 and approve purchases, backed by new gateway and seller tools.

At a glance
- Walmart announced AI agent shopping controls on Oct. 21, with Agent Profiles launching within a month.
- Shoppers can set agent spending limits such as $50, $100 or $200, or approve purchases one at a time.
- Only 11% of 1,000 U.S. adults surveyed by NMI in August had used AI to complete a transaction; 70% wanted review or override rights.
- The Agent Gateway, built on Anthropic's Model Context Protocol, will launch in limited beta for companies and developers.
- Vitamins and protein supplements were the top two Walmart product types purchased through ChatGPT, per Daniel Danker.
Walmart will launch Agent Profiles within the next month, letting customers approve AI agent purchases, set spending limits of $50, $100 or $200, and track every agent-placed order. The retailer announced the tools Wednesday, Oct. 21, across Walmart and Sam's Club.
The move targets a real trust gap. In an August survey by payments technology provider NMI of 1,000 U.S. adults, only 11% had used AI to complete a transaction. Another 70% wanted the ability to review or override an AI purchasing decision before checkout. Just 10% would give AI full control over a purchase.
Walmart acknowledged that not everyone will shop through AI agents. For those who do, the company said its goal is to "make our experience the most trusted" by building safeguards into the shopping process.
What are the three new tools?
Walmart is introducing:
- Agent Profiles: Shoppers authorize AI agents, set spending limits and track purchases.
- Agent Gateway: A more direct connection between agents and Walmart's shopping services.
- New purchasing rules: Limits on agents' actions and access to customer information.
The changes extend Walmart's existing AI shopping efforts. Those include the Sparky AI shopping agent and integrations with Meta's Muse, OpenAI ChatGPT and Google's Gemini. Walmart has not specified which agents will support the new profiles at launch.
How do Agent Profiles work?
Today, purchases made by AI agents appear alongside customers' own orders with no way to tell them apart. The new feature fixes that. Agents will "shop using dedicated profiles linked to a customer's account, so customers can approve purchases, set spending limits, and see which purchases were made by an agent," Daniel Danker, Walmart's executive vice president of AI acceleration, product and design, said in a LinkedIn post.
Setup takes three steps:
- Start in the agent's app: Walmart creates a profile when a customer first shops through its Agent Gateway.
- Connect to Walmart: Customers sign in to their Walmart accounts to authorize and link the agent's profile.
- Set spending controls: Shoppers choose a limit, review agent purchases and can delete profiles at any time.
Shoppers can approve AI purchases one at a time or allow spending up to a preset cap. On Walmart's website, agents will see "the same prices that a human would see," with offers and fees displayed before purchase.
What protections apply when agents shop?
Walmart is "setting expectations for how AI agents work with customers and our company," Danker said. The new terms require customer approval of agent purchases by default, though Walmart said it could still require additional approval even within preset limits.
The rules bar agents from acting beyond a shopper's instructions, accessing unauthorized information or receiving payment-card details. Agents cannot add or change payment methods. Unintended purchases fall under Walmart's existing returns process, and agents that violate the terms will be blocked from the website. Information shoppers share directly with third-party AI apps remains subject to those providers' privacy policies.
The Agent Gateway itself is built on Model Context Protocol, the open standard developed by Anthropic that lets AI applications connect with external tools and services. Today, agents browse Walmart using steps designed for people. "This can be slow and error-prone, and agents may get stuck along the way," Walmart said. The gateway supports "discovery, add to cart, purchasing, and order tracking," Danker said, and it will launch in limited beta for companies and developers. Walmart has not announced timing or participants.
The gateway can also help an agent find a customer's preferred brand of milk or bread "without sharing the customer's broader shopping history."
What does this mean for marketplace sellers?
Agentic commerce is already reshaping product discovery. At Morgan Stanley's Technology, Media and Telecom Conference in March, Danker said vitamin and protein supplements were the top two product types purchased from Walmart through ChatGPT — purchases that may have begun as broader conversations rather than product searches.
That shift is a "potential game-changer" for sellers, said Manish Joneja, senior vice president of Walmart Global Marketplace and Walmart Fulfillment Services. Agentic commerce "could create more opportunities for sellers to put the right products in front of the right customers, including brands and products that may not have broad awareness," Joneja said.
The bar rises too. Joneja said accurate product information, competitive prices, availability and reliable delivery promises will all become increasingly important. Walmart is working with sellers to improve listings and catalog quality so products are "easier for AI-powered shopping experiences to understand and recommend." Walmart ranks No. 2 in Digital Commerce 360's Top 1000 Database and No. 37 in its AI Commerce Rankings — a gap the retailer's agent infrastructure is built to close.
Original: subscription.digitalcommerce360.com
More from James Calloway
Show full bio
Correspondent covering technology and product news in trade marketing, shopper marketing, trade promotions, category management at Target Marketing.
77 articles