Meta's Muse: a 200-person, seven-month bet on personal AI agents
Meta Chief AI Officer Alexandr Wang said in an Oct. 11 interview with David Senra that Muse, the personal AI agent Meta launched in September, now has millions of users and may be one of the fastest-growing consumer AI apps since ChatGPT — though he did not disclose retention, inference-cost or revenue figures. Muse went from prototype to launch in seven months with a team of fewer than 200 people; the team drew up more than 100 required behaviors and capabilities for an ideal personal agent, built evaluations and release red lines for each, and held the product until the specially trained Muse Spark 1.3 was the first version to clear all of them. Wang said the effort began with intensive use of OpenClaw by himself and Meta VP Nat Friedman, and that Muse is meant to act for users — booking, searching, communicating and arranging deliveries — as trust grows.
The interview ties Meta's consumer-agent strategy to a reliability bar and a deliberately small team, offering a lens on whether its heavy AI spending can turn into user growth.