Meta's Muse AI Agent Hits No. 1 App Store, Passing ChatGPT

Ten days from launch to the top of the chart

Meta's Muse, a personal AI agent launched on September 8 across iOS, Android, the web at muse.ai, and WhatsApp, reached No. 1 on Apple's U.S. free iPhone chart on September 18 — ten days after launch — with roughly 730,000 U.S. downloads over that stretch, according to Sensor Tower estimates15. In taking the top spot, it moved ahead of OpenAI's ChatGPT, Google's Gemini, Anthropic's Claude, and Meta's own Instagram18. Meta has attributed the climb to new installs and rapid feature rollouts, including a Mac app that arrived on September 1749.

The natural comparison is ChatGPT's App Store debut in 2023, and the numbers invite two opposite headlines. ChatGPT recorded about 697,000 U.S. downloads in its first eight days, per data.ai, yet at that point it sat at only No. 3 on the U.S. free chart, per an Appfigures snapshot115. Muse hit No. 1 with roughly 730,000 downloads in ten days — a slightly larger total, over a longer window, in a chart environment where the top slot is apparently easier to claim519. Meanwhile, Muse's daily pace of about 73,000 U.S. downloads trails ChatGPT's 2023 average of roughly 87,000 per day over its first eight days1519. So the honest reading is both: Muse edged ChatGPT's cumulative launch figure while lagging its velocity, and it seized a No. 1 rank that ChatGPT never held at a comparable point. Some outlets compressed the timeline — one report described the top ranking as arriving "just a week" after launch, though the launch date and the September 18 peak put it at ten days29.

The numbers under the ranking are thin

Beneath the chart position, the early data is modest by Meta's own standards. Muse opened with roughly 83,000 U.S. iOS downloads in its first two days — about 41,000 per day — before accelerating in week two59. For context, Meta's Threads pulled more than 4.3 million U.S. downloads on its launch day, and the Meta AI app drew 108,000 in its debut111217. Muse also opened more slowly than ChatGPT did in 2023: ChatGPT passed half a million U.S. installs in under a week, a run rate it took Muse roughly twice as long to approach121317. Android traction has been particularly weak, with Muse sitting at No. 338 in Google Play's Productivity category early in its run, and download figures there not yet available1314.

That backdrop matters because chart rank measures velocity, not sustained use. The Mac Observer cautions that the No. 1 position is a snapshot that can shift as downloads move, and that users should examine permissions and connected-service access before handing an agent a task10. A rank built on a compressed marketing burst says more about paid distribution than about whether people keep using the app a month later — the metric no download tracker can yet supply.

Agent, not assistant: what Muse actually does

The product itself is the more consequential story. Where ChatGPT began as a chatbot that answers questions, Muse is built to complete tasks — sending emails, booking travel and dinners, and making purchases on a user's behalf once connected to their apps and personal accounts18. The App Store listing spells out the scope: it manages scheduling conflicts, sets reminders and follows up with people, audits subscriptions, flags forgotten charges, cancels what users don't need, and builds workout or meal plans from data users choose to share, checking in before it cancels anything or spends money6. The app currently shows about 24,000 ratings averaging 4.9 and ranks No. 1 in Productivity, with an 18+ age rating and free download with in-app purchases6.

The agent-versus-assistant framing is deliberate. One guide to the app draws the line cleanly: Meta AI is the assistant that answers; Muse is the agent with its own secure computer21.

The AI browser angle: an agent with its own machine

The most technically interesting piece is how Muse gets things done. It runs on what Meta calls Muse Secure VM — a dedicated cloud computer with its own web browser — so it can search the web, fill out forms, book appointments, and make purchases through connected services2122. Crucially, it keeps working after the user closes the app, and it asks for approval before sensitive actions such as sending an email or buying something21.

This is the direction consumer AI is heading: not a chat window inside an app, but an autonomous browser session in the cloud acting on your behalf. The smartphone becomes a remote control for a machine that clicks through airline checkout pages and subscription cancellation flows that no one wants to navigate by hand. Meta's launch framing for the agent is deliberately broad — an assistant that "gets things done across every part of life," from finances and health to shopping22. If agents with their own browsers become the default interface for errands, the battleground shifts from model quality to how many services an agent can reliably operate — a fight where owning distribution channels and logged-in user accounts counts for a lot.

How Meta manufactured the climb

The rise to No. 1 was engineered as much as earned. The rollout began with a closed alpha and a waitlist with invite codes on September 7, followed by the full U.S. launch the next day and Meta's first ads by September 9320. Ratings jumped overnight from 422 to 2,270 between September 12 and 13, a Mac app landed on September 17, and September 18 brought both a Canada launch and a fresh wave of advertising — the same day Muse claimed No. 1320. The app is reportedly all over the App Store's paid placements, and analysts credit a combined playbook of waitlist, referral tokens, rating prompts, a mascot, and heavy ad spend executed simultaneously320.

One skeptic's framing cuts through the hype: it's tempting to credit Meta's billions of users and deep pockets, but that explains less than it appears, since Muse reached No. 1 on a daily download rate roughly half of ChatGPT's 2023 debut pace320. What the money bought was the top slot on thin volume — a demonstration of marketing muscle, not a landslide of organic demand.

The trust problem nobody's solved

The brake on agent adoption isn't capability — it's permission. Muse connects to email, calendars, payment methods, and other personal accounts, then acts on the user's behalf8. A survey of 1,500 U.S. consumers found only 8% said they would trust Meta with their passwords8. That single number explains why Muse's launch pace is softer than its ranking: the app isn't asking users to try a novelty chatbot, it's asking them to hand over the keys to their digital lives. The App Store listing leans hard on reassurance — approval before spending or canceling, data users "choose to share" — because that friction is the product's biggest liability621.

The verdict: a category arrives, the race begins

Strip away the chart optics and the story is straightforward: consumer appetite for true task-doing agents is real but unproven, Meta is willing to spend whatever it takes to plant its flag first, and the ChatGPT comparison flatters Muse in one dimension while undercutting it in another. Muse hit No. 1 because the free chart rewards a concentrated burst of installs in a soft week, not because it matched ChatGPT's cultural moment in 2023 — when an AI app topping the charts was itself the news115.

What matters next is retention. ChatGPT turned downloads into daily habits and a durable App Store presence that Muse just displaced for the first time. Muse's real test is whether users who connected their email and payment cards in week one keep the connections live in month three — and whether Meta can extend beyond a U.S.-iOS-skewed base, given weak Android traction and a Canada launch that only began September 1831314. For now, the agent era has its first chart-topping app, won on downloads that trail the bot era's benchmark. The rank is Meta's; the category, and the trust deficit that comes with it, is still very much up for grabs.