Less than three weeks after launch, Meta's Muse has become the most downloaded free app in the United States, overtaking ChatGPT and turning the personal AI agent from a developer curiosity into a mass-market product. Meta is now pouring its advertising machine, its hardware ambitions and its Connect keynote into the app, betting that consumers want an assistant that acts on their behalf rather than one that simply answers questions.
From Launch to No. 1 in Ten Days
Muse launched on September 8 in the US and Canada, built on Meta's Muse Spark family of models. Meta pitches it as a way to direct capable digital assistants that can complete tasks across the web, such as filling out online forms and organising email inboxes. Early users have been pointing it at chores like hunting for deals, booking reservations and comparing insurance policies.
The uptake has been fast by any measure:
- 902,000+ downloads in its first six days, according to Sensor Tower data cited by Bloomberg, ahead of the roughly 773,000 the Meta AI app managed over the same post-launch window.
- No. 1 on Apple's US App Store on September 18, followed by the top spot on Google Play the next day.
- An estimated 2.8 million downloads in its first two weeks, with average day-over-day growth of 55%, compared with about 24% for ChatGPT in its first ten days, per Sensor Tower.
- More than 3.4 million downloads by Sensor Tower's latest count through Thursday, though other trackers disagree: Apptopia estimates about 4.3 million and Appfigures roughly 2.3 million.
The spread between estimates is a reminder that third-party app data is directional rather than precise. But every tracker agrees on the trend: Muse is growing faster than any consumer AI launch in recent memory.
Meta Puts Its Muscle Behind Muse
Part of that growth is paid for. Meta began promoting Muse across Facebook and Instagram on September 9, and within ten days the app had absorbed most of the company's house-ad inventory, crowding out promotions for Facebook, Instagram and WhatsApp themselves. Sensor Tower said Muse ranked among the top ten brands by ad spend by September 22, as TechCrunch reported.
At Meta Connect on September 23, the company expanded the product considerably. Announcements included:
- Avatar video chat with the Muse agent.
- Mac support, letting the agent work on desktop computers.
- A dedicated email address for each user's agent.
- Plans to integrate Muse with Meta's smart glasses.
- Muse Charm, a small Tamagotchi-like device that shows an animated avatar of the agent and lets users talk to it after tapping a fingerprint sensor.
Mark Zuckerberg said Muse Charm will ship in time for the December holidays, though Meta has not disclosed a price, battery life or launch countries. Daily active users rose 27% the day after the keynote, according to reports on Sensor Tower data.
On monetisation, Zuckerberg said the app will be free for what he described as "a huge number of tokens," with Meta eventually taking a small fee on transactions the agent completes. That model positions Muse less as a subscription chatbot and more as a commerce layer, sitting between consumers and the businesses they buy from.
Growing Pains: Security and Platform Friction
Rapid adoption has brought scrutiny. Security researcher Patrick Wardle found that Muse's macOS app could be manipulated to redirect its server connection and steal a token granting full control of the agent. Meta patched the flaw quickly, but the episode underlined how much damage a compromised agent with access to inboxes, accounts and payments could do.
Muse has also run into resistance from other platforms. Amazon moved to block the agent from its store earlier this month, an early sign that retailers may not welcome third-party agents shopping on their sites without an agreement in place.
Why It Matters
Muse is the first agent product to reach genuine mass-market scale, and it arrives in a week dominated by headlines about rogue AI agents breaching systems they were never meant to touch. That combination will shape how regulators, retailers and security teams think about consumer agents.
For the industry, three signals stand out:
- Distribution beats model quality. Meta's ability to push an app to billions of Facebook and Instagram users gives it a head start that standalone AI labs cannot easily match.
- The business model is shifting to transactions. If agents earn fees on purchases, the fight over which agent sits in front of the consumer becomes a fight over e-commerce margins.
- Security is now a consumer issue. A stolen agent token is closer to a stolen wallet than a leaked password, and consumer agent apps will need bank-grade safeguards.
Whether Muse keeps its lead once Meta's promotional blitz slows remains an open question. For now, it has shown that ordinary users will hand real tasks to an AI agent when the product is placed right in front of them.
