Meta’s AI assistant is now a shopping channel. Does it recommend your business?

Meta launched Muse, a personal AI agent, on September 8. It works in its own app and in WhatsApp. On Meta’s third-quarter earnings call on October 21, 2026, Mark Zuckerberg and CFO Susan Li gave the first usage numbers:

• Sign-ups: “In just a month and a half, more than 10 million people in the US and Canada have set up Muse personal agents” (Zuckerberg).
• Daily use: “About 25% of people who have set up Muse use it on any given day” (Zuckerberg).
• Shopping comes first: “The most common use case for Muse is shopping” (Zuckerberg). Li put it at “over 25% of Muse weekly actives”.
• Large purchases: “an average purchase size over $200” (Zuckerberg). Asked whether that includes travel, he said it does, and that travel planning is another top use.
• Demand beyond forecasts: “in the first week we had to work with Stripe to raise credit limits since the volumes were above what any of us expected” (Zuckerberg).
• Small businesses: Meta estimates “at least 500 million self-identified small businesses” use its apps, and says Muse “could help these businesses acquire customers” (Li).

These are Meta’s own figures, from the call transcript on its investor site.

A search engine gives a shopper ten links, and the shopper picks one. An agent like Muse does the picking. Li described it on the follow-up call: Muse is “doing proactive research for you, researching, providing recommendations, researching alternatives, helping you complete a purchase.” An agent “can compare relevant options, it can answer questions, it can help them complete the purchase.”

When the shopper only sees a shortlist, a business the agent couldn’t read or didn’t trust never gets compared. A business whose facts it got wrong may be compared on the wrong price, hours or service area. And shopping isn’t limited to online stores. Zuckerberg said the purchase figure includes travel, and Meta’s launch post lists booking travel among Muse’s tasks. Restaurants, clinics, contractors and hotels sit in the same answers as retailers.

Meta hasn’t published how Muse chooses what to recommend. What it has said:

• Muse has its own browser. It “runs on Muse Secure VM, a dedicated secure computer with its own browser”, and “it can open a browser, fill out forms” for the user.
• Meta reads websites for Meta AI. Meta’s crawler documentation says that allowing Meta-WebIndexer in robots.txt “helps us cite and link to your content in Meta AI’s responses”. Meta-ExternalFetcher fetches pages a user asks for, including “helping AI navigate websites to complete tasks for users”, and may bypass robots.txt.
• Product catalogs feed it. Shopify says products from its stores “are shared with Meta by default through Shopify Catalog”. Shoppers can also check out inside Muse, and Shopify says Shop Pay “can display as a wallet or checkout option for customers in Muse”.
• Agents will talk to agents. Meta previewed a Personal Agent Protocol with Sierra, Walmart, Stripe and Shopify. It defines how a shopper’s agent talks to a business’s agent. Li said “many experiences will remain AI assisted rather than fully autonomous” in the near term.

In practice, most local and service businesses will be found through what Meta’s systems can read on the web: your own site and the listings and pages that describe you.

1. Let Meta’s crawlers in. Check that your robots.txt doesn’t block Meta-WebIndexer, Meta-ExternalAgent or Meta-ExternalFetcher. A web firewall’s bot challenge or a broken HTTPS certificate can also stop an agent’s browser even when robots.txt allows it. The free AI crawler check reads your robots.txt and homepage the way AI crawlers do.
2. Put the facts an agent compares in plain text. That means what you sell or do, prices or a price range, service area, hours, booking or ordering steps, and return or cancellation policy. Put them on your own pages as text, not only in images or PDFs.
3. Give every product or service its own page. An agent comparing options needs a page it can open for each one.
4. If you sell on Shopify, check the Meta channel. It’s under Sales channels > Agentic > Meta AI and Muse (Agentic Storefronts) in your Shopify admin. Decide whether you want catalog sharing and direct checkout on, rather than leaving the default unexamined.
5. Keep your facts identical everywhere. Your website, Google Business Profile, Facebook and Instagram pages, Yelp and industry directories should agree. When they don’t, AI contradicts your website, and a shopper may never check which one is right. See why AI gets business hours wrong.
6. Ask the questions your customers ask. Open Muse and ask the questions without your business name, such as “best family dentist near [your town]” or “where to buy [your product] under $200”. Note who it names and what it says about you.

Preferly’s $99 report asks 10 AI models, including Meta Muse Spark, the model Muse runs on by default, what they tell customers about your business. It shows where AI contradicts your website and gives the fixes. It checks what the models say in their answers. It doesn’t test the Muse app’s shopping, its product catalog or its checkout, and AI agent results, including Muse, are coming soon. The free AI crawler check needs no email. It checks whether 14 AI crawlers, Meta’s included, may read your site.