Meta’s Muse personal AI assistant gains traction as consumer AI gets harder

Meta has a consumer AI hit on its hands, at least for now. Muse, the company’s personal AI agent, launched on September 8, 2026, and pulled in approximately 2.5 million downloads in its first 13 days.

What Muse actually does

Meta is pitching Muse as more than a chatbot. The company describes it as a proactive assistant that pursues goals and completes tasks, instead of waiting for someone to type a question.

Under the hood, Muse runs on the latest versions of Meta’s Muse Spark model family. It operates inside a proprietary secure virtual machine, which works like a sealed-off digital workspace where the agent can act without touching the rest of your device.

The agent connects to a range of user applications. It is available on iOS, Android, and the web.

The numbers so far

Muse recorded more than 900,000 downloads in its first week, then climbed to approximately 2.5 million by day 13.

In mid-September, Muse passed ChatGPT in the US iOS free app rankings.

Meta’s stock rose more than 20% in the weeks after Muse launched.

Meta also offers paid tiers. The Power plan costs $20/month, and the Maximum plan costs $100/month. Most users are expected to stick with the free tier.

Commerce enters the picture

At its Connect event on September 23, Meta announced expanded merchant support. Brands including Walmart and Best Buy are part of that push.

The trust problem

Muse works best when it can see your stuff. To get full functionality, users have to grant access to their personal accounts. That requirement has raised privacy concerns and trust issues among consumers.

The secure virtual machine is part of Meta’s answer. By running tasks in an isolated environment, the company is trying to show that granting access doesn’t mean handing over the keys to everything.