Meta launched Muse on September 8, a personal AI agent that browses the web, fills out forms, and completes purchases on a user's behalf. It runs on iOS, Android, muse.ai, and WhatsApp in the US, with a free tier plus $20 and $100 monthly plans.

The part that matters for sellers isn't the shopping capability. It's that Muse has its own browser and checks out on external merchant sites, rather than routing through a partner catalog the way most agentic commerce has worked so far.

How the Checkout Actually Works

Payment runs through Link, Stripe's consumer wallet. At more than one million merchants that already accept Link, Muse checks out instantly using the payment method a shopper has saved. Everywhere else, Link issues a single-use virtual card scoped to that specific approved purchase, so the agent never touches the buyer's real card details.

Every transaction requires the user to confirm the total in the chat interface before payment proceeds. Meta says Muse is the first AI agent covered by Link's purchase protections, which include coverage for damaged or lost items, price-drop refunds, no-fee returns, and a return guarantee on eligible purchases. Shop Pay and 1Password support are planned but weren't available at launch, and Meta gave no dates.

The Attribution Problem Nobody Has Answered

Here's what sellers should be paying attention to. Muse can browse websites, complete forms, and check out on participating sites, potentially removing several steps where publishers and affiliate platforms would ordinarily record customer activity. Meta has not announced how affiliate links, publisher influence, or product-source attribution get handled.

That's a real gap. If an agent navigates directly to your product page and completes a purchase, the referral path that would normally tell you a customer came from a creator's affiliate link, a comparison site, or a paid ad may simply not exist in your data. You get the order. You may not get the story of where it came from.

This is a different problem than the one Shopify's catalog-based approach creates, where at least the traffic is identifiable as AI-referred. An agent operating its own browser looks more like direct traffic than anything else.

How This Differs From the Other Agentic Commerce Plays

Three major approaches are now live, and the differences matter for how you'd prepare for each.

Anthropic released blueprints letting retailers build their own branded shopping agents, with checkout staying in the retailer's system. Amazon built Alexa for Shopping inside its own ecosystem, where it controls discovery end to end. Meta's Muse does neither. It operates independently across the open web, using a payment layer that works whether or not a merchant has integrated anything.

That last part is why the reach number is significant. Muse doesn't need merchant participation to buy from a site. Link's single-use card handles the checkout regardless.

One Claim Worth Reading Carefully

Meta stated that Muse does not share a person's conversations or virtual machine data with Meta's ad systems. Given that Meta's business is almost entirely advertising, walling off what would be an extraordinarily rich personal data stream is a meaningful commitment.

It's also unaudited and stated in a press release. Worth tracking rather than assuming.

What to Do Now

Nothing about Muse requires action from you today. It's US-only, brand new, and there's no merchant program, advertiser tooling, or partnership structure attached to it.

The useful preparation is the kind that pays off regardless. Clean structured product data, crawlable prices, plain-text policies, and a checkout that doesn't depend on unusual interactions all improve conversion for human shoppers now and make your site readable to agents later. That's the same groundwork that improves AI discovery on Shopify and Amazon, so it isn't a speculative investment.

On attribution specifically, start watching for unexplained direct traffic that converts unusually well. If agentic purchasing scales, that's where it will show up first in your analytics, and knowing your baseline now makes the shift easier to spot later.