New Sensor Tower data shows Meta’s AI agent bringing back 73% of new users the next day, nearly double ChatGPT’s rate. The harder tests are whether people keep coming back once the novelty fades, and whether the stores Muse shops in will let it through the door.
Meta’s new AI agent is pulling people back faster than its biggest rival. According to Sensor Tower data reported by Yahoo Finance on Monday, 73% of people who install Muse open it again the next day, nearly twice ChatGPT’s rate. Investors noticed quickly, sending Meta shares up about 4% to roughly $745 by Monday afternoon, with Yahoo Finance calling Muse the main driver of the stock. Muse launched in September and had passed 5.1 million downloads by early October, according to Sensor Tower figures cited by TipRanks, while Similarweb estimated 2.18 million daily users on Oct. 5.
What sets Muse apart is that it is built to do things, not just answer questions. Deutsche Bank analyst Benjamin Black describes it as a mobile app with a consumer-friendly interface that can complete transactions on its own, runs continuously in its own virtual machine and is mostly free to use. That combination puts Muse in a different lane from a chatbot, because it is designed to book, buy and handle errands for the person holding the phone. Black estimates Muse could account for as much as 8% of Meta’s revenue by 2030, roughly $36 billion, with a low-end estimate of $2.4 billion.
The first-day number is the headline, but it is not the whole story. Muse loses users faster after that first return visit, and ChatGPT overtakes it between days eight and nine, according to the same Sensor Tower data. Muse still holds second place, which is a strong debut for a product barely a month old. The pattern is a familiar one in consumer technology, where curiosity brings people back once and habit is what keeps them.
The bigger constraint may be access. Major retailers have begun blocking AI agents from browsing their sites or completing purchases, with Amazon the leading example, and Amazon stayed out when Meta and Sierra launched an open standard for how AI agents shop on your behalf, even as Shopify, Stripe and Walmart signed on. An agent built to buy things is only as useful as the stores that let it in, and the country’s biggest online store currently has its door closed.
That fight matters well beyond Meta’s share price. If agents become the way many people shop, book travel and pay bills, the companies that decide which agents get access will shape prices, choices and which businesses get seen at all. Small merchants that never join an agent standard could become harder to find, and shoppers who lean on a free agent will see whatever that agent’s owner has arranged to show them. Muse is free today, but Meta’s business has always been built on advertising, and every revenue forecast for Muse assumes the money comes from somewhere.
For now, Muse has shown that people will try an AI that acts on their behalf and return to it the next morning. It has not yet shown that they will make it a habit, or that the open web will make room for it. The next round of retention data, and whether Amazon changes course, will say more about Muse’s future than one good day on the stock market.
Sources: Yahoo Finance (Sensor Tower, Similarweb and Deutsche Bank data); TipRanks; 24/7 Wall St.