On September 8, 2026, Meta launched Muse, which it calls “the world’s first personal AI agent built for everyone.” Ten days later, it was the #1 free app on the US iPhone App Store — and it held the spot for nearly two weeks. Analysts estimate around 3.4 million downloads in a single week, with cumulative US downloads passing 5 million by the end of September.
The difference from every chatbot you’ve tried: Muse doesn’t just answer questions. You give it a goal — restock the household supplies, plan a trip, chase an unpaid invoice — and it works through its own secure virtual machine with its own browser to get it done, checking back with you before anything sensitive. Meta says Muse won’t publish content, send messages, or spend money unless you authorize it.
What Muse actually does
Under the hood is Muse Spark, Meta’s language model (also sold to developers through its Model API). The agent runs tasks autonomously: researching options, comparing prices, filling forms, booking, and completing multi-step errands across the web.
Commerce is wired in from day one. Meta’s launch materials cite integrations with Shopify and Shop Pay, Expedia, PayPal, and Stripe — whose executive called Muse the first AI agent covered by Link’s purchase protections. That matters: an agent that can buy things needs payment rails people trust, and Meta lined them up before launch.
The pricing is aggressive: free for up to 100 million tokens a week, Power at $20/month (500 million tokens a week), and Maximum at $100/month (3 billion tokens a week). It launched in the US on September 8, expanded to Canada on September 18, shipped a Mac app on September 17, and runs on iOS, Android, and the web. No Windows app yet — desktop beyond Mac means the browser.
The enterprise play
On September 28, Meta went further, announcing the Meta Enterprise Platform — a dedicated business unit led by former MongoDB CEO CJ Desai, reporting directly to Mark Zuckerberg. The launch lineup includes four products: the Muse agent itself, Meta Business Agent (for WhatsApp, Messenger, and Instagram), the Muse API for developers, and Muse Code, a terminal and CI coding agent.
This is Meta’s serious push into enterprise AI monetization. Analysts at Citi project the platform could generate $27 billion in annual revenue by 2030, split between transactions and subscriptions. Wells Fargo raised its Meta price target to $1,000, citing Muse and AI-driven product cycles.
For context on how seriously Meta is taking this: the company reportedly cut internal access to Anthropic’s Claude Code from around 60,000 to 30,000 users, after spending topped $105 million in 28 days — and is building its own customer-facing coding model instead.
Muse vs. the rivals
The agent race is now a four-way sprint. Google launched its Gemini agent for enterprise in the same window, with “coworker agents” that get their own identities and email addresses. OpenAI shipped Dots, always-on agents with their own cloud computers, at DevDay. Meta’s edge is distribution: billions of people already live inside WhatsApp, Instagram, and Messenger. If the agent meets users where they already are, Meta doesn’t need to win the model benchmark — it needs to win the default.
What this means for you
If you run a small business: the Meta Business Agent angle is the one to watch. An agent living inside WhatsApp, Messenger, and Instagram — where your customers already message you — handling inquiries, bookings, quotes, and follow-ups could replace a meaningful chunk of manual customer communication. Add it to your vendor watchlist now and request access early; early adopters of business messaging tools historically get the cheapest customer-acquisition costs.
If you sell online: Muse’s commerce integrations (Shopify, Shop Pay, Stripe) point to a near future where customers delegate purchases to agents. The practical move today: make your product listings clean, structured, and agent-readable — clear titles, accurate specs, real reviews. The businesses AI agents can understand will be the ones they buy from.
If you’re a freelancer: delegation-style agents are a brand-new service category. “I’ll set up your Muse workflows” — connecting the agent to your client’s store, calendar, and inbox — is a consulting offer that didn’t exist two months ago. Learn the platform before your competitors do.
If you’re a developer: the Muse API (Muse Spark at $1.25 per million input tokens and $4.25 per million output tokens) is a competitively priced option for building agent features into your own products.
One honest caveat
A privacy dispute flared around launch — Meta denies Muse accessed a user’s messages without consent — and repeat-use data isn’t public yet. Downloads are not loyal users; the real test is whether people delegate their second, fifth, and fiftieth task. And with the agent able to reach ad accounts and payment systems, the authorization settings deserve your full attention before you connect anything financial.
Still, the direction is unmistakable: AI is moving from conversation to delegation, and Meta just put a free, capable agent in front of more people in ten days than most AI products reach in a year.
Sources: Meta’s official newsroom and muse.ai — Muse launched September 8, 2026; Meta Enterprise Platform announced September 28, 2026.
Keep Reading
Keep exploring with these related HustlePrompt guides: