Agentic Commerce: Walmart Opens the Door, Amazon Builds a Wall

Two of the world's largest retailers just answered the same question in opposite ways. Should brands let artificial intelligence agents shop on behalf of customers, or should they keep the algorithms out entirely? Walmart's answer, delivered this week through its new Agent Profiles and Agent Gateway, is a resounding yes, complete with spending caps and approval flows. Amazon's answer, delivered a month earlier by blocking Meta's Muse agent, is a hard no.

This split is not a footnote in retail tech. It is the first major stress test of agentic commerce, the emerging system in which AI agents research, compare, and complete purchases for consumers with minimal human clicking involved. Bain forecasts the US agentic commerce market will reach $300-500 billion by 2030 and account for 15-25% of total eCommerce sales , and Gartner projects that 20% of digital commerce transactions will be executed through AI platforms, either via on-platform checkout or via AI agents by 2030 . That is not a distant trend line. It is a business model decision executives need to make now.

Matt Britton, founder of consumer insights platform Suzy and author of Generation AI, has spent the better part of two decades tracking how emerging technology reshapes buying behavior before Wall Street notices. He argues this moment deserves the same urgency boardrooms gave to mobile commerce in 2010 and social commerce in 2015. "The brands treating this as a feature update are going to get outflanked by the brands treating it as a channel shift," is the type of framing Britton brings to Fortune 500 keynote audiences grappling with where to place their bets.

This piece breaks down what Walmart actually built, why Amazon chose the opposite path, and what both moves signal about the next era of consumer trust. It also outlines the practical steps business leaders should take before agentic commerce becomes table stakes rather than a competitive edge.

What Is Agentic Commerce and Why Walmart's Move Changes the Calculus

Agentic commerce refers to transactions initiated, negotiated, or completed by an AI agent acting on a consumer's behalf rather than the consumer navigating a site or app directly. Instead of a shopper typing a search query and clicking "add to cart," an agent like ChatGPT, Claude, or Meta's Muse interprets intent, compares options, and executes the purchase within rules the person has set. It is the natural extension of voice assistants and chatbots, except now the agent can actually spend money.

Consumer adoption is already well underway. Research from IBM and the National Retail Federation found 45% of consumers already turn to AI for part of their buying journey , and Kearney data shows 73% of consumers are familiar with AI tools and 60% expect to use AI agents for shopping within the next 12 months . Walmart read those numbers and concluded the only responsible move was to build the infrastructure before someone else's agent showed up uninvited.

That is precisely what was happening before this announcement. Today, an agent shopping at Walmart signs in with the customer's own account, so its purchases are mixed in with the customer's , leaving no clean way to track, audit, or limit what the agent actually bought. Walmart's new framework fixes that by giving every agent its own identity inside the customer relationship, a move Britton frequently discusses on his Speed of Culture podcast as the defining shift separating reactive retailers from proactive ones.

Inside Walmart's Agent Profiles and Agent Gateway: How Spending Caps Actually Work

Walmart's announcement centers on two new systems working together. Agent Profiles let shoppers authorize AI agents, set spending limits and track purchases, while the Agent Gateway gives agents a more direct connection to Walmart's shopping services . A third piece, new purchasing rules, sets limits on agents' actions and access to customer information .

The mechanics matter for any executive studying this as a blueprint. Here is how the system is structured:

• Dedicated identity: Agent Profiles, due within the next month, give each agent a separate profile inside that account , so agent purchases never blend into a shopper's regular order history.
• Flexible approval modes: Through Agent Profiles, shoppers can choose to approve AI purchases one at a time, or they can allow agents to spend up to a preset limit, such as $50, $100 or $200 .
• Revocable access: Customers can delete a profile at any time , giving them a kill switch if an agent misbehaves.
• Zero payment exposure: Walmart's new terms bar AI agents from seeing card details or acting beyond what shoppers intend .
• Fair pricing guarantee: On Walmart's website, agents will see the same prices that a human would see, with offers and fees displayed before purchase .

Walmart itself offered the clearest analogy for how this works. The retailer described the model by noting, "Think about it like adding someone as a cardholder on your credit card" . That framing is deliberate and smart. It borrows decades of consumer trust built into financial services and applies it to software, a pattern Britton has flagged repeatedly when advising clients in the financial services sector about how trust frameworks transfer across industries.

Under the hood, the Agent Gateway runs on the Model Context Protocol, an open standard that lets approved agents query inventory and execute checkout without scraping Walmart's site like a human browser would. Agents now shop on Walmart's site through a web browser, following steps designed for people, and Walmart called that slow and error-prone, so the Walmart Agent Gateway instead connects agents through the Model Context Protocol, an open standard . In practice, Walmart is not just tolerating AI agents. It is building dedicated plumbing to make them faster and more reliable than human checkout ever was.

Amazon's Walled Garden: Why Blocking Meta's Muse Tells a Different Story

Amazon has taken the opposite approach, and it has not been subtle about it. The retailer blocked Meta's new agent, Muse, from shopping its site, and that single enforcement action sits inside a much longer pattern. Amazon's website rules block many AI assistants, including ChatGPT, Claude, Copilot, Perplexity, Grok, and Google crawlers , while the company simultaneously promotes its own Alexa shopping features.

Amazon's stated rationale focuses on consent and identity. The company said "Continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed" when users attempted to shop through Muse. Amazon also claims it wasn't aware in advance that Muse would access its store, and did not authorize it , and argues that using an AI agent for shopping bypasses the personalization and insights the company has built into the shopping experience .

The commercial logic underneath that explanation is harder to miss. Amazon's advertising business is enormous, and agents that skip the browsing experience skip the ads entirely. Amazon's advertising business, worth more than $68 billion last year, depends on people browsing its pages, and an agent that shops on a person's behalf can skip them . That single data point explains more about Amazon's strategy than any public statement the company has issued.

Industry analysts see the bind Amazon has created for itself. As one retail technology observer put it, "Amazon now faces a paradox. If it stays closed it risks ceding the AI-agent layer to competitors. But if it opens up, it hands control of the customer relationship to an external platform" . Britton frames this tension as the central dilemma of agentic commerce: every retailer must now choose between protecting a legacy revenue model or capturing a new distribution channel, and very few companies can credibly do both at once.

Marketing to Algorithms: The New Battleground for Brand Visibility

Walmart's open approach creates a reality that most marketing teams have not yet priced into their strategy. If an AI agent is the one comparing products, reading reviews, and making the purchase decision, then the agent, not the human, becomes the primary audience for product listings, descriptions, and pricing signals. Brands that have spent years optimizing for human search behavior now need a parallel strategy optimized for how large language models parse and rank product data.

This is not a hypothetical shift. Consumer behavior data shows the funnel has already started moving upstream into AI tools. Among shoppers who use AI for purchasing decisions, 63% compare brands, models, prices and reviews, 55% use it to learn about a category or product, and 46% use it for inspiration and discovery . Every one of those moments is now an opportunity for an algorithm, not a human, to decide whether a brand even makes the shortlist.

Britton often tells audiences at his AI keynote presentations that this is the clearest evidence yet that marketing departments need a second discipline sitting alongside SEO: answer engine optimization for the agents themselves. Structured product data, transparent pricing, and clean inventory feeds are no longer back-office concerns. They are now frontline marketing assets that determine whether an agent recommends a product at all.

Enterprise adoption of agentic systems is accelerating fast enough that this shift cannot wait for a slow rollout. Gartner's forecast puts that number at 67% by end of 2026 and above 85% by 2028 for enterprise deployment of agentic AI, meaning the question for most large companies is no longer whether to prepare, but how quickly they can catch up. Britton's keynote work, detailed on his speaker platform, increasingly centers on helping leadership teams build that readiness before competitors lock in the first-mover advantage.

Trust and Control UX: Why Spending Caps Are the New Consumer Loyalty Currency

Walmart's decision to lead with spending limits and approval flows, rather than raw AI capability, reveals something important about where consumer trust actually comes from in this new environment. Shoppers are not primarily worried about whether an agent is smart enough to pick the right product. They are worried about whether it can be trusted not to overspend, misuse data, or act outside its mandate. An agent can make decisions within the permissions a shopper grants, and researcher concerns about autonomous shopping systems are relevant because AI agents have been used against shopping websites in other contexts, which is why Walmart's profile and approval model is intended to place boundaries around legitimate use .

That is why the control mechanisms Walmart built, not the AI itself, are the real headline. Shoppers can require approval on every single purchase, or set a hard dollar ceiling, and they retain the ability to revoke access instantly. This gives Walmart a trust advantage that competitors without comparable guardrails simply cannot match right now.

Britton has long argued, including in research conducted through Suzy, his consumer insights platform, that younger shoppers in particular demand transparency and control before they hand over autonomy to any brand or technology. That pattern is especially visible among Gen Z consumers, a group Britton covers extensively in his Gen Z keynote speaker engagements, who have grown up expecting granular permission settings on every app they use. Retailers that fail to offer that same granularity in agentic commerce risk losing a generation of shoppers who simply will not hand an algorithm a blank check.

The stakes extend well beyond retail. Industries built on recurring high-trust transactions, from real estate closings to mortgage approvals, are watching this playbook closely, which is part of why Britton's work increasingly touches sectors like the one covered on his real estate AI speaking page. Any industry where a customer hands over financial decision-making authority to software will eventually need its own version of Walmart's approval flow.

Key Takeaways for Business Leaders

• Audit your current e-commerce infrastructure to determine whether AI agents can already access your storefront unofficially, the way they did on Walmart's site before Agent Profiles launched.
• Decide deliberately whether your brand's long-term advantage sits in openness, like Walmart, or in protecting a proprietary customer relationship, like Amazon, rather than drifting into a default position.
• Build structured, agent-readable product data now, since algorithms will increasingly determine which brands appear in comparison shopping before a human ever sees the result.
• Design trust and control features, including spending limits and revocable permissions, as core product requirements rather than legal afterthoughts.
• Train marketing and e-commerce teams to treat answer engine optimization with the same urgency once reserved for traditional search engine optimization.

Frequently Asked Questions About Agentic Commerce

What is agentic commerce and how is it different from online shopping?

Agentic commerce is the process by which an AI agent researches, compares, and completes a purchase on a consumer's behalf, rather than the person clicking through a website themselves. It differs from traditional online shopping because the agent interprets intent and executes transactions within permissions the consumer sets in advance, shifting the primary "user" of a retail site from a human to a piece of software.

How do Walmart's Agent Profiles protect consumers from overspending?

Walmart's Agent Profiles require customer approval before an agent completes a purchase by default, and customers can optionally set a spending cap such as $50, $100, or $200 that the agent cannot exceed without additional approval. Walmart's new terms require customers to approve agent purchases by default, though shoppers can authorize spending within preset limits . Customers can also delete an agent's profile instantly if it behaves unexpectedly.

Why did Amazon block Meta's Muse AI shopping agent?

Amazon blocked Muse after accusing Meta of accessing its platform without authorization and without the agent properly identifying itself as AI software while browsing. Amazon blocked Meta's Muse AI agent from accessing its store, saying Meta lacked permission , and the company has applied similar restrictions to other outside AI agents, including ChatGPT and Claude, to protect its advertising-driven browsing experience.

Should my company open its storefront to AI shopping agents?

The right answer depends on whether your competitive advantage comes from controlling the full customer journey, like Amazon, or from maximizing reach and transaction volume, like Walmart. Companies with strong proprietary customer data and advertising revenue tied to browsing behavior may lean toward restriction, while companies prioritizing growth and distribution should consider building agent-friendly infrastructure with clear trust and spending controls now.

How fast is consumer adoption of AI shopping agents growing?

Adoption is accelerating quickly across research and purchasing behavior alike. Kearney reports that 73% of consumers are familiar with AI tools and 60% expect to use AI agents for shopping within the next 12 months , signaling that agentic commerce is moving from novelty to mainstream expectation within a single year.

Walmart and Amazon just drew the two opposing lines that every retailer, bank, and consumer brand will eventually have to choose between. One company is betting that trust, transparency, and openness will win the agentic era. The other is betting that control over the customer relationship matters more than short-term reach. Matt Britton has built his career identifying these inflection points before they become obvious to the broader market, and agentic commerce is shaping up to be one of the defining ones of this decade.

Business leaders who wait for more certainty will likely be reacting to decisions their competitors already made. Those who want a clear-eyed, data-driven roadmap for navigating agentic commerce, AI shopping agents, and the broader consumer trends reshaping retail should explore Matt Britton's AI keynote speaker programs or review his full speaking platform for Fortune 500 engagements. The agents are already shopping. The only remaining question is whether your brand built the storefront they can trust.