Meta Muse for brands: what the AI agent reads and does.

Muse is Meta’s personal AI agent, launched in the US on 8 September 2026, and it works for the person using it, not for the brand they follow.

That distinction matters because Muse Spark 1.1, the model underneath it, is often described as a social media planning and scheduling tool. Meta has never said that. Muse Spark 1.1 is a reasoning model, released on 9 July 2026, and nothing Meta has published says it plans, posts or schedules content for a brand account.

For a brand, the useful questions are narrower and more commercial: what Muse can read from a customer’s Instagram, what it cannot do on your behalf, where it is available, and what a UK brand selling to US customers should check now. We cannot use Muse from the UK, so this is not a hands-on review. Everything below comes from Meta’s own announcements and help pages, and from Shopify’s documentation, with the dates they carry.

What Muse actually is

Meta describes Muse as a personal AI agent that takes on tasks and goals rather than just answering questions. Its examples are ordinary errands: sending emails, booking travel, building a plan for a goal and working through it.

It runs inside what Meta calls Muse Secure VM, a dedicated virtual machine with its own browser. From there it can open web pages, fill in forms and, in Meta’s word, negotiate. For longer tasks it keeps working after the app is closed and comes back when it needs approval.

People talk to it like messaging a contact, in the Muse app or directly in WhatsApp. Meta says Muse checks with the person before sensitive actions such as sending an email or making a purchase.

Muse Spark, Muse Spark 1.1 and Muse Image

Three names get blurred together, so here is how they fit.

• Muse Spark is the model family. Meta calls it its most capable model to date, built for real-world agentic work, and dates its first announcement to April 2026.
• Muse Spark 1.1, announced on 9 July 2026, is a multimodal reasoning model with gains in tool use, computer use, coding and understanding images. Meta made it available in Thinking mode in the Meta AI app and on meta.ai, and to developers through the Meta Model API in public preview.
• Muse Image is Meta’s image generation model, which rolled out in Meta AI on 7 July 2026, with more than 30 AI effects in Instagram Stories.

We covered what Muse Image means for content teams separately. The agent is the newer and bigger story for brands, because it acts rather than generates.

What Connect 2026 added

At Connect on 24 September 2026, Meta announced more for Muse, with most of it described as coming soon rather than shipped:

• Muse on Meta’s AI glasses, acting on what the wearer is looking at.
• A voice mode for long conversations while Muse works in the background.
• Its own email address, which Muse uses to complete tasks and correspond.
• More commerce and work connectors, including Walmart, Best Buy, Sephora, Ulta, Wayfair, Gap, Michael Kors, Shop Pay, PayPal, Expedia, Instacart, Notion and GitHub.
• Muse Charm, a pocket device for talking to Muse, with details promised later this year.

Meta put the connectors in “the coming months” without dates or countries. Treat that list as announced, not live.

What Muse reads from Instagram, Facebook and Threads

This is the part brands should read twice. According to Meta’s help centre, Facebook, Instagram and Threads connect to Muse automatically when they are linked through a person’s Accounts Centre, and the person can manage those connections there at any time.

Meta’s own launch example shows what that means in practice: Muse can turn a recipe reel someone saved on Instagram into a grocery list, and suggest a menu for their dinner party.

Read that as a buyer journey. A saved reel is no longer only a signal to the feed. It can become an instruction to an agent that shops, compares and fills in checkout forms. Meta also says connectors exchange only the data a task needs, and that Muse remembers what matters to a person until told to forget it.

What Muse does not do for a brand account

No Meta page we have seen says Muse posts, schedules or replies on Instagram, Facebook or WhatsApp Business on a brand’s behalf. It is a consumer agent working for its user.

So Muse does not replace a content calendar, an approvals process or a community manager, and it does not change who is accountable for what your account publishes. If your team is being sold “Muse for social” as a workflow product, ask the seller to show you the Meta page that says so.

The operational shift is on the other side of the screen. Your content is increasingly being read by software acting for a customer, as well as by the customer.

Where Muse is available, and when

• United States: launched 8 September 2026 on iOS, Android and muse.ai.
• Canada: Meta said on its X account on 18 September 2026 that Muse had reached Canada, on iOS and the web first, with Android to follow.
• UK and EU: Meta has given no date. Its subscription page says only that Muse “isn’t available in all locations yet”.

Muse is free for most uses, with paid plans at $20 and $100 a month. Subscribers must be 18 or the age of majority where they live.

For UK teams, the honest position is to wait for Meta’s own announcement. Do not route staff through a VPN to test it: a workaround tells you nothing reliable about how Muse behaves for a real US customer.

The merchant side: Shopify and the connectors

The commerce plumbing is live already for some sellers. Shopify’s help centre says direct checkout on Meta is activated by default for eligible stores, and that product discovery is powered by Shopify Catalogue, which feeds Muse (where available) and other Meta surfaces.

Eligibility is narrow: you must sell to customers in the United States, Canada or Mexico, and Meta’s direct checkout shows only to customers in those countries. Stores need their terms, privacy and refund policies in place.

Opting out takes two switches under Sales channels, then Agentic, then Meta: turn off Shopify Catalogue access to stop sharing products, or turn off direct checkout to keep purchases on your own store. For a UK brand shipping to the US, this may already be switched on.

That matters for the Facebook and Instagram side of a brand as well as the store. If the product an agent finds does not match what your reels and captions promise, the customer notices after checkout, not before.

Muse phoning your business

Meta says it is experimenting with letting Muse make phone calls on a user’s behalf to businesses whose phone numbers are publicly available, for example to ask about opening hours or make a booking.

Meta’s help centre offers businesses an opt-out form. It does not say which countries the test covers or how many businesses have been called. If your public number reaches a shop floor, a salon or a reception desk, decide whether you want AI calls, and brief the people who answer the phone either way.

Muse and Meta’s ad systems

Meta states that Muse does not share a person’s conversations or the data in their virtual machine with Meta’s ad systems, and that people can opt out of their interactions being used to train Meta’s AI.

For brands, the practical reading is simple. On Meta’s word, what a customer tells Muse does not become targeting anyone can buy. What an agent can weigh up is what it can read about you in the open: your content, your catalogue and your product pages. That is organic work.

What a UK brand with US customers should check

1. Your Shopify settings, if you sell to the US, Canada or Mexico: is Catalogue access on, and is direct checkout on? Decide deliberately.
2. Your product data: titles, prices, sizes, stock and returns policy, written so a machine can read them without guessing.
3. Your saved and shared content: do your reels name the product, the price and where to buy it clearly enough to become a list or a basket?
4. Your public phone numbers: whether to use Meta’s opt-out form, and what your staff say to an AI caller.
5. Your reporting: a sale that starts with a saved reel and ends through an agent will not look like a normal referral, so expect gaps before you trust any attribution.

How NBK thinks about agents and social

Agents like Muse reward the same things good organic social always has: clear packaging, consistent product information and content people save because it is useful. We have argued the same about how AI is changing creative strategy: the tools move fast, and the system behind the content decides whether you benefit.

The difference now is that some of your audience is software. It reads literally, it compares ruthlessly and it does not forgive a caption that promises something your product page contradicts.

Next step

If your content and your catalogue tell different stories, an agent will find the gap before a customer does. NBK can help you audit the system behind your social, so what you publish holds up when it is read by a person or by the agent working for them.