Shopify Opens to Meta Muse; Amazon Balks

Shopify is welcoming Meta’s new Muse personal AI agent. Amazon is trying to keep it out.

Meta introduced Muse this month as a personal AI agent that can browse websites, fill out forms, and complete purchases with a user’s consent. Muse’s virtual machine, or Muse VM, gives the agent its own browser environment for carrying out these tasks.

What’s more, Muse, like many agents, can remember interactions and use them in tasks.

Shopify added Meta’s Muse as an AI commerce channel, while Amazon has blocked Muse from its marketplace. The opposing responses show how unsettled agentic commerce remains.

Meta’s Muse can browse websites, fill out forms, and complete purchases with a user’s consent. Image: Meta.

Shopify

Shopify added Meta to its Agentic Storefronts program on September 8 (the same day Muse launched).

Shopify shares merchant products with Meta by default, although sellers can manage catalog access and direct checkout. Shopify’s agentic settings let merchants view Meta performance.

An ecommerce platform, such as Shopify, may have an incentive to treat agentic commerce as just another distribution channel.

The platform earns revenue when its merchants sell, regardless of whether the shopper arrives through a traditional storefront, social platform, search engine, or AI agent.

If Muse can discover products via an ecommerce integration and consummate transactions, Shopify and other platforms can preserve their role in catalog, checkout, payments, order management, and merchant infrastructure even if the agent controls more of the customer-facing experience.

Amazon

Amazon appears to have taken an opposite approach.

Muse users attempting to shop on Amazon now see a message stating “continued access by an unauthorized AI agent violates Amazon’s Conditions of Use.” In published reports, Amazon said Meta had not obtained authorization before allowing Muse to access the marketplace.

This should not be a surprise. As a marketplace, Amazon must own the shopping interface, the advertising environment, and much of the customer relationship.

An outside AI agent such as Muse could bypass parts of that experience, reducing the marketplace’s control over product discovery and retail media, and inserting a bot between the marketplace and the shopper.

2 Responses

Both positions make sense. An AI agent could become a discovery tool, a sales channel, a shopping interface, or an intermediary.

Consider how agentic commerce might change just three areas.

Storefronts

For the most part, an online store in 2026 is both a customer-facing experience and the infrastructure to complete a transaction.

Shoppers engage with an ecommerce website, and behind the scenes the platform acts as a content management system and checkout software.

With agentic commerce, however, a shopper might tell Muse to find a gift for his niece for less than $100 that will arrive by Friday. The agent could research products, read reviews, compare stores, check availability, and recommend a gift without the shopper ever navigating a retailer’s website.

An ecommerce platform might still provide the product data, but the agent, not the storefront, would control the shopping interface and experience.

Advertising

Advertising could change, too.

Google, Amazon, retail media networks, affiliates, and publishers all benefit from reaching motivated shoppers.

A persistent AI agent could learn about that intent before the shopper reaches traditional advertising environments.

For example, a consumer might tell Muse she needs a carry-on bag for a month in Italy, has a $250 budget, and prefers real leather. The agent could begin comparing products before the shopper visits Google, Amazon, or a merchant.

In that case, the agent may decide which products enter consideration before a shopper sees a sponsored search ad, retail media placement, or display ad.

Recommended

The Race to Own AI Shopping

September 6, 2026

Knowledge of customers

An AI agent’s persistent knowledge of customers could have an even larger impact.

Retailers have spent years trying to assemble a useful view of individual customers from transactions, website behavior, loyalty programs, email engagement, and other first-party data.

Now, Muse and similar agents can remember information a user provides and apply it to tasks. In a shopping context, that persistent knowledge could include sizes, preferred brands, budgets, delivery expectations, and purchases.

Thus, instead of every retailer rebuilding a partial customer profile, the agent could become the keeper of the shopper’s preferences and intent, carrying that knowledge from one buying experience to the next.

The company controlling the agent could therefore influence more than product discovery. It could also control an increasingly important layer of the customer relationship.

It is a race to own AI shopping.

High Stakes

Shopify appears to be betting that agents such as Muse can become another route to a sale. Amazon insists those agents enter its marketplace on its terms.

Agentic commerce is evolving, but the two opposing responses suggest the stakes are high.