Meta says more than 10 million people have set up Muse, and about a quarter use the AI agent daily. Shopping is among its leading uses, with more than a quarter of engaged shoppers averaging purchases above $200.
Those figures, disclosed on Meta’s October 21, 2026 third-quarter earnings call and reported by TechCrunch, explain the company’s commercial ambition: keep Muse free for most people, take a percentage of commerce and charge power users for additional AI capacity.
The opportunity is consequential. An agent that helps people buy things could earn money from transactions, giving it a different business model from a subscription chatbot. But Meta’s figures describe different populations and time horizons. Separating them shows what the company has demonstrated through its reported usage, and what remains a forecast.
What the 10 Million User Figure Measures
Meta’s adoption claims cover account setup and daily activity. They are useful signals, but they answer different questions.
| Reported metric | What it tells readers | What remains unknown |
|---|---|---|
| More than 10 million people set up Muse | The product has reached a substantial audience | How many remain active over longer periods |
| About 25% use it daily | A meaningful share has incorporated it into everyday activity | How that share changes as individual groups of users age |
| More than a quarter of engaged shoppers average purchases above $200 | Some shoppers make relatively large purchases | Shopper numbers, purchase frequency and total spending |
Applying roughly 25% to a base of 10 million suggests daily usage on the order of 2.5 million people. That is a calculation from Meta’s rounded figures, not a separately disclosed daily-active-user count.
Zuckerberg presented daily usage as helping retention. Measuring retention directly would require following the same groups over time: how many people who joined in one month still use Muse a month or three months later? A snapshot of daily activity cannot answer that.
These are company claims conveyed through TechCrunch’s report. The available evidence does not include the earnings transcript or underlying usage measurements.
