Recent media reports have highlighted Amazon’s decision to block Meta’s Muse AI agent from making purchases on its marketplace. Muse has quickly gained attention for its ability to research products, complete tasks, and shop on behalf of consumers, while Amazon appears focused on maintaining direct control of the customer relationship and checkout experience.  

The situation is interesting, but retailers should resist the temptation to view it as a definitive signal about the future of commerce.  

At the moment, we see this as two technology giants competing in a quickly evolving commerce landscape. Meta wants Muse to become a gateway to digital experiences, including shopping. Meta has tried many times to be a destination for commerce (e.g., Facebook stores, Instagram shopping). Amazon wants to remain the starting point for product discovery and purchase – and it has billions of dollars in e–commerce, marketplace dollars, and ad revenue at stake.  

It’s worth remembering that relationships among major technology firms are rarely static. Google pays Apple for valuable real estate on iPhone screens, and – who knows? – it would not be surprising to perhaps see Amazon and Meta eventually strike some form of commercial arrangement if agentic shopping gains meaningful traction. 

For now, digital leaders should bear in mind that: 

  • It’s still unclear whether a critical mass of consumers actually want autonomous purchasing. Agentic commerce generates excitement, but shopper behavior suggests a more measured future. Even relatively lightweight automation such as subscription replenishment programs has seen limited adoption.
  • Specialized use cases might be where we see meaningful adoption. Agentic shopping may prove valuable during high-intent events like Prime Day or during holiday promotions. Consumers might embrace automation to grab a limited edition item just as it drops, to secure concert tickets in the best section when they go on sale, or to pursue a refund on an item they didn’t receive. That doesn’t mean that most consumers will give AI broad authority to spend money without clear guardrails, approval flows, and transparency. Today, most adoption will be among early adopters and those who aren’t concerned with permissions and access issues with AI agents.
  • Many processes are best handled within merchants’ systems. For example, the post-transaction shipment and returns flow includes logic and communications that the merchant needs to control. Neither Meta nor any other third party can execute on the entire process, though an AI agent might be handy to initiate a return on behalf of a customer.
  • The scary stories about these agents might delay adoption. Tech news is bursting with stories about adverse outcomes for users, from an AI agent booking a hotel and joining its rewards program on behalf of a user without permission, or an AI agent selling an item on Facebook Marketplace for well below the asking price and then arranging a pickup meeting without informing the seller. Some early adopters may simply back away from their use, while others will delay adoption entirely until the agents are more stable. 

In the near term, the strongest use case may be research rather than purchasing. This use case is something that we – including my colleagues Emily Pfeiffer, Chuck Gahun, and I – have been advising repeatedly, and as Emily and Chuck did in this Forrester webinar. AI agents can help consumers discover products, compare options, and narrow choices. Converting recommendations into transactions is still a much bigger leap for most consumers. 

If you’re a Forrester client and would like to discuss these questions further, please book a Guidance Session with us.

Share