Meta’s Muse has expanded a beta that lets its personal AI agent place outbound calls to businesses in the United States, TechCrunch reports. Meta is initially giving access to people who had asked Muse to make a call, turning a feature request into part of the rollout queue.
The update matters because the current crop of personal agents is trying to move beyond chat windows and into chores that still require a person, a browser or a telephone. Meta launched Muse earlier this month as an agent for tasks such as sending email, booking travel and selling a car, according to WIRED’s launch reporting. Meta says the service runs through its own apps, the web and WhatsApp, with access and task permissions at the center of the product’s pitch.
Calling is becoming a table-stakes feature
Muse is not alone in adding the capability. Instinct announced Instinct Concierge, an early-access feature its founder said can handle jobs including restaurant bookings, dentist cancellation lists and cable-bill issues. TechCrunch reported that both products added calling in the same window, bringing them closer to parity on a feature that other task-agent services had used to stand out.
That does not mean the products work the same way. Instinct framed Concierge as a limited early-access service, while Meta’s initial Muse beta is for U.S. businesses and prioritizes people who had already requested calling. The reports do not establish broader availability, call success rates or how either service handles a difficult real-world conversation. Those details will matter more than a launch checklist once agents are routinely speaking for users.
Early enthusiasm is not a product verdict
One public post praised Muse as an unexpectedly strong Meta product after the author’s negative view of earlier Meta chatbots and Llama 4. That reaction is useful as a snapshot of the launch conversation, but it is not an independent test and does not establish that Muse led the category on calling.
The more concrete signal is that calling has quickly become a shared ambition. TechCrunch also cited Sensor Tower data showing more than 730,000 U.S. Muse downloads in its first five days, compared with 707,000 for the Meta AI app over the equivalent period. Download interest is not proof of retention or reliable automation, but it helps explain why rivals are racing to make their assistants do more than answer a prompt.