BREAKING
AI Tools

Walmart Lets AI Agents Shop for You — With Spending Limits Built In

Updated October 22, 2026
10 min read
Editor's Note: This article was human-reviewed for accuracy
Woman using a smartphone while pushing a shopping cart full of groceries in a supermarket aisle

Quick Answer & Key Takeaways

  • • Walmart announced Agent Profiles and a dedicated Agent Gateway on October 21, 2026, letting AI agents like ChatGPT, Gemini and Meta's Muse search and buy at Walmart and Sam's Club.
  • • Shoppers get a separate profile per agent, per-transaction approvals or preset spending limits, and agents never see payment card data.
  • • Purchases cost nothing extra, require no Walmart+ membership, and Walmart's standard returns apply even when an agent misbehaves.

On Wednesday, October 21, 2026, the world's largest retailer by revenue picked the exact opposite strategy from its biggest rival. While Amazon has spent the past two months blocking AI shopping agents from its platform, Walmart CEO John Furner stood up at the Masters of Scale Summit in San Francisco and announced Walmart would build them a front door instead — with spending limits, per-agent receipts, and a written rulebook for how agents must behave.

This isn't a vague partnership teaser. It's three concrete pieces of infrastructure: Agent Profiles that give each AI agent its own identity inside your account, a Walmart Agent Gateway that lets approved agents talk to Walmart's systems directly instead of clumsily driving a browser, and a formal set of terms agents must follow or get blocked. Here's what each one does, what it costs, and why Amazon's refusal to do the same makes this the defining retail story of the AI agent era.

The quick answer

  • What launched: Agent Profiles (arriving next month), the Walmart Agent Gateway (in beta for companies and developers now), and formal terms for AI agents — all announced October 21, 2026.
  • How you stay in control: each AI agent gets a dedicated profile inside your Walmart account; you can require approval for every purchase or let it buy within a spending limit you set (Walmart's examples: $50, $100, $200); you can delete the profile at any time.
  • What agents can do: search Walmart and Sam's Club inventory and complete purchases through a direct MCP connection, without navigating the website like a human.
  • What they can't do: see your payment card data, add or change payment methods, or act beyond what you intended — and Walmart's standard returns apply even when an agent messes up.
  • What it costs: nothing extra. No Walmart+ membership required — any Walmart account holder can use it.

What Walmart actually announced

Today, an AI agent that shops on Walmart's site logs in with your credentials and does a passable impression of a human clicking through pages built for people. Walmart's announcement calls that what it is: slow and error-prone. The three new tools replace it with something designed for software.

1. Agent Profiles — a separate identity for every agent

Launching "in the next month" — so November 2026 — Agent Profiles give each AI agent a dedicated profile inside your Walmart account instead of letting it borrow your login. You authorize the profile, link it to your account, and every purchase the agent makes appears on its own separate ledger, apart from your own orders.

The controls come in two flavors. You can require your approval for every purchase, or you can let the agent buy freely within a spending limit you set — Walmart's own examples were $50, $100, and $200. The setup starts in the agent's app: the first time you use an agent to shop through the Gateway, Walmart creates its Agent Profile automatically; you then authorize and link it, set your limit, and you can delete the profile any time to cut the agent off instantly.

Walmart's analogy is the best explanation of the product: it's like adding someone as an authorized user on your credit card. They get to spend, you see every transaction, you cap the limit, and you can revoke it. Only here the "someone" is a piece of software that shops at 3 a.m.

2. The Walmart Agent Gateway — stop driving the browser

The Gateway is an MCP server — built on the Model Context Protocol that Anthropic originally created — giving approved agents a direct, reliable pipe into Walmart's inventory and checkout instead of scraping and clicking a website built for humans. It starts as a beta for companies and developers, and Walmart lists Meta's Muse, ChatGPT, and Google's Gemini among the agents that will connect through it.

Here's the part that's cleverer than it looks. The Gateway also solves the privacy problem: it lets Walmart resolve an agent's "get my usual stuff" request against its own data without handing the agent your shopping history. Walmart's example is almost homely — if you always buy a particular sourdough bread or a specific brand of milk, Walmart finds those items for the agent without sharing your broader purchase history or personal information. The agent gets the answer, not the dataset.

One honesty note from the fine print: Walmart's own beta page calls this an early, invite-only beta — access is limited, it "may be suspended at any time," and "some features may change and may not be available to all participants." This is a foundation being poured, not a finished floor.

3. Terms for AI agents — the rulebook

Approval is required by default, and Walmart reserves the right to demand additional approvals even for purchases inside your spending limit. Agents must not act beyond what the shopper intends. They never receive payment-card information and can't add or change payment methods. They can access only the information the task actually needs.

And the clause that does the most for consumer confidence: Walmart's standard returns process applies — including when an agent doesn't act as intended. If your agent orders the wrong thing, you're not stuck negotiating with a robot's mistake. Walmart also updated its Terms of Use (last revised September 21, 2026) to define agents explicitly — software that accesses Walmart's sites on a user's behalf, including AI assistants and browser-automation tools — and to state that anything an authorized agent does within its granted permissions counts as the customer's own action. Break the rules, and the agent gets blocked from the site.

The pricing fine print: nothing costs extra

These tools are free for shoppers. Walmart says agent support will be available to anyone with a Walmart account — no Walmart+ membership required. There's no tier, no upsell, no agent-access fee disclosed anywhere in the announcement.

The pricing commitment that actually matters is subtler: Walmart says it prices products, not people, so an AI agent sees the same Every Day Low Price a human would see on Walmart.com, with applicable offers and fees disclosed before the purchase completes. One of the open anxieties of agentic commerce is whether algorithms will quietly get different — worse — prices than humans. Walmart has now gone on record that they won't.

Why this matters: Amazon just did the opposite

The contrast is the story. Over the past two months, Amazon has gone adversarial on agentic shopping: it blocked Meta's Muse from its site in September, showing users a notice that unauthorized AI agents violate Amazon's terms of use, and it blocked Perplexity's shopping agent as well. Amazon's answer is first-party: Alexa for Shopping and its Buy for Me feature, on Amazon's terms. TechCrunch adds that travel and booking sites have taken the same door-slamming approach, locking agents out of their websites completely.

Walmart, meanwhile, was already the retailer most wired into the agent ecosystem — a partnership with Meta's Muse, conversational-commerce integrations with ChatGPT and Gemini, plus its own Sparky assistant. This launch turns that presence into infrastructure.

The industry is noticing. Mizuho analyst David Bellinger wrote that Walmart's new tools instill confidence in what he called the "experimental" process of agentic purchases, adding that "the momentum and buy-in from top management teams seems to be making this into a real thing." And the momentum on the agent side is genuinely real: Meta's Muse has been downloaded more than 10 million times, according to third-party data analyzed by Sensor Tower; Anthropic launched its proactive Claude Duet agent on the very same day as Walmart's announcement; OpenAI launched always-on Dots agents in September. The agents exist and they're shopping. Walmart's bet is that shoppers will prefer the store that lets them in over the one that bolts the door.

The honest pros and cons for shoppers

What's genuinely good here. The separate per-agent ledger fixes the real annoyance of today's setup — agent purchases mixed invisibly into your own history. The spending limits are the right default control: most people will set a $50 or $100 cap for a grocery run rather than approve every transaction by hand. The no-payment-data exposure is non-negotiable done right — the agent never touches your card. And the returns guarantee is the single feature that makes early adoption sane; without it, letting software spend your money would be a leap of faith.

What to be realistic about. The Gateway is an invite-only beta that can change or be suspended — this is day one, not a mature product. Only approved agents get through the front door; everyone else is still clicking the website or getting blocked. The approval-by-default setting is safe but will feel tedious until you trust the limits enough to switch. And the deepest risk isn't technical at all: do you actually want your grocery decisions delegated to software whose idea of "your usual milk" comes from a model? Walmart kept that data to itself for a reason — but the judgment call stays yours.

Walmart vs Amazon: the two bets on AI shopping

WalmartAmazon
Third-party AI agentsWelcomed — official Agent Gateway, written terms, spending controlsBlocked — unauthorized agents violate terms of use
Per-agent spending limitsYes — Agent Profiles with per-purchase approval or set limitsNo equivalent for third-party agents
Payment data exposureAgents never receive card data; can't add or change payment methodsN/A
Returns on agent mistakesStandard returns apply, including when the agent misbehavesN/A
First-party AI shoppingSparky assistant; integrations with Muse, ChatGPT, GeminiAlexa for Shopping, Buy for Me

Frequently asked questions

When do Walmart Agent Profiles launch?

Walmart said "in the next month" at the October 21, 2026 announcement, which points to November 2026. The Agent Gateway itself is already in beta for companies and developers, but that beta is invite-only.

Which AI agents work with Walmart's Agent Gateway?

Walmart names Meta's Muse, OpenAI's ChatGPT, and Google's Gemini among the agents that will connect through the Gateway, alongside developers building through the beta program. Only approved agents get the direct MCP connection.

Can the AI agent see my credit card?

No. Under Walmart's terms, agents never receive customers' payment-card information and cannot add or change a customer's payment methods. They can only access the information needed for the task you've authorized.

Do I need a Walmart+ membership to let an agent shop for me?

No. Walmart says agent support will be available to anyone with a Walmart account — a Walmart+ membership is not required.

What happens if the agent orders the wrong item?

Walmart's standard returns process applies, including when the agent doesn't act as intended. That's written into the agent terms, and it's the safety net that makes the whole thing viable for early adopters.

Key takeaways

  • On October 21, 2026, Walmart announced Agent Profiles, the Walmart Agent Gateway, and formal terms for AI agents — three pieces that make agentic shopping an official, controlled feature at the world's largest retailer by revenue.
  • Agent Profiles (coming in November 2026) give each AI agent its own profile inside your account, with per-purchase approval or a preset spending limit — deletable at any time.
  • The Agent Gateway is an MCP server giving approved agents like ChatGPT, Gemini, and Meta's Muse a direct, reliable connection to Walmart and Sam's Club systems — no browser-driving, no scraping, and no wholesale access to your shopping history.
  • Agents never see payment card data, can't change payment methods, and standard returns apply even when agents misbehave; agents that break the terms get blocked.
  • Amazon is going the exact opposite direction — blocking third-party agents — making this a straight Walmart-versus-Amazon bet on how agentic commerce should work.

As of October 22, 2026, the AI shopping race is really a race over trust: Walmart is writing the rulebook and letting agents in, while Amazon is bolting the door. Watch November's Agent Profiles rollout — how many agents get approved, and how quickly shoppers actually hand one a spending limit, will decide which bet was right.

Sources

10 min read
5 views
edit_note Written by

Ratul Ahamed

Founder & Editor-in-Chief

Ratul Ahamed is the Founder and Editor-in-Chief of AI Core Updates, steering the publication's strategic growth, technological architecture, and human-verified journalistic standards. With over eight years of hands-on experience across full-stack systems and machine learning infrastructure, Ratul founded AI Core Updates to deliver transparent, benchmark-backed intelligence that bridges the divide between foundational research and real-world executive decision-making. At AI Core Updates, he directs long-term editorial investigations and leads the development of internal automated verification workflows.

Read all stories by Ratul Ahamed arrow_forward
verified_user Human-Reviewed Editorial

forum Comments (0)

Leave a Reply

chat_bubble_outline

No comments yet. Be the first to join the conversation!

AI Core Updates Insider Dispatch

Stay Ahead of the Exponential AI Curve

Join over 48,000+ AI researchers, software engineers, and founders. Get our daily 5-minute morning intelligence briefing analyzing breakthrough research, frontier models, and autonomous workflows.

bolt
Daily Briefing in 5 Mins
High signal, zero marketing fluff.
podcasts
Podcast Deep Dives
Early access & episode transcripts.
psychology
Evaluated Prompts & Tools
Tested benchmarks & system designs.
verified_user
Strict Anti-Spam Policy
1-click unsubscribe at any time.
AK
SL
MR
JD
Trusted by builders from leading AI labs & enterprises.
mark_email_unread

Subscribe to Newsletter

Delivered daily at 7:00 AM UTC

person_outline
mail_outline
Include in my subscription:

No spam ever. You can easily unsubscribe with a single click at any time.

cookie

Cookie Settings

We use cookies to enhance your browsing experience, serve personalized content, and analyze our traffic. By clicking "Accept All", you consent to our use of cookies. Learn more