Meta's Muse can now call U.S. businesses for users, but only paid subscribers

Meta's consumer AI assistant Muse can now place phone calls to U.S. businesses on behalf of its users, according to TechCrunch. The capability, which Meta first tested in September, moved from a limited experiment to broad availability on November 12 — with one significant catch: only paying subscribers can use it.

From limited test to paid feature

Muse has passed 10 million users, a figure Meta shared last month, according to TechCrunch, and the assistant is positioned as the company's attempt to turn consumer AI into an everyday utility. Handling phone calls fits that ambition by taking a widely disliked chore off people's to-do lists.

Alexandr Wang, Meta's chief AI officer and head of its Superintelligence Labs division, announced the launch in a social media post, pointing to insurance companies, cable providers, the DMV and dinner reservations as examples of calls the agent can take over. During the earlier testing phase, only a small set of early adopters could try the feature, and they had to request access directly from the assistant. Now any paying customer can ask Muse to place a call.

What access costs

According to TechCrunch, Muse subscriptions start at $20 per month for 500 million Muse tokens per week, while heavier users can pay $100 per month for 3 billion tokens. Call placement sits behind that paywall for the time being, making it one of the clearest cases yet of Meta reserving a specific agent capability for revenue-paying customers rather than the free tier.

The free tier is not frozen, however. Wang said this week that Muse's recently added email address option, which lets the agent manage correspondence, will soon reach non-paying users, TechCrunch reports.

A mixed reception on the other end of the line

Muse is not the first agent to pick up a phone. TechCrunch notes that the startup Instinct offers a comparable service, as does Google's Gemini. The harder problem is what happens once someone answers. The report describes the reception as mixed: businesses and receptionists often hang up as soon as they realize the caller is not human. At the same time, synthetic voices have become convincing enough that distinguishing an agent from the customer it represents is getting harder — a dynamic that cuts both ways, helping agents finish calls while making organizations more suspicious of their incoming lines.

Why it matters

The launch is a small but concrete milestone for AI agents: a major platform's consumer assistant can now act in the world's coordination layer — the telephone system — on a paying customer's behalf, at the scale of a 10-million-plus user base. It also targets one of the most durable consumer frustrations in software-adjacent life. As TechCrunch observes, customers have spent years working through phone menus, waiting on hold and being bounced between departments; an agent that will sit on hold indefinitely is a credible substitute.

Two open questions follow. The first is commercial: whether enough subscribers find the feature worth the subscription, and whether Meta eventually extends calling to free users as it is doing with email. The second is social: businesses now face inbound calls placed by software, and the etiquette, disclosure norms and screening practices for that traffic are still being improvised. One agent booking a table is a convenience; large volumes of agents dialing government offices and customer-service lines is an operational question that has not been fully answered.