Meta Launches Business Calling Feature for Paid Muse AI Subscribers in the US
Meta has expanded the capabilities of its consumer artificial intelligence agent Muse, enabling the software to place autonomous telephone calls to United States businesses on behalf of users, according to TechCrunch. Access to the automated calling functionality is restricted exclusively to paying subscribers across the United States following preliminary testing initiated in September. Meta reported last month that Muse has surpassed 10 million active users, representing a key effort to establish consumer artificial intelligence as an everyday utility. Under current pricing tiers, entry-level Muse subscriptions begin at $20 per month for 500 million weekly tokens, while heavy users can access 3 billion tokens for $100 per month.
Practical Applications and Everyday Utility
The calling service is designed to manage tedious telephone interactions, including scheduling appointments, navigating automated menu trees, or contacting venues lacking digital booking options. Alexandr Wang, Meta's Chief AI Officer and head of its Superintelligence Labs division, stated that the tool targets administrative chores that consumers frequently avoid, such as contacting cable providers or municipal motor vehicle registries. The automated agent handles dialling, enduring hold times, and conversing directly with receptionists to extract details or complete transactions. Wang noted that while call placement remains paywalled, an associated feature allowing users to establish dedicated Muse email addresses will expand to non-paying users. Consumers frustrated by traditional customer service delays can utilize the software to navigate automated phone trees and endure hold queues on their behalf. After each call, the Muse user receives a transcript and a summary, according to Layer3 Labs.
Privacy Safeguards and Autonomous Architecture
The commercial rollout follows an earlier experimental trial that incorporated human contractors to assist the software when autonomous dialogues faltered. Meta discontinued that contractor-assisted model after user privacy concerns emerged regarding third-party listeners monitoring private consumer phone calls. The company subsequently committed to a fully synthetic voice infrastructure, deploying automated disclaimers to ensure transparency during external business calls. The voice deployment operates on hardware clusters running 12 concurrent real-time voice sessions per Nvidia GB200 processor at an operational latency of roughly 870 milliseconds. According to a Reuters report cited by Layer3 Labs, internal tests found that human-placed calls succeeded 95% to 98% of the time, whereas AI-only calls performed worse as some businesses hung up upon detecting an artificial voice.
Commercial Reception and Competitive Landscape
Meta faces practical operational hurdles on the line, as commercial receptionists and enterprise telephone systems frequently disconnect upon recognizing synthetic machine voices. However, improvements in conversational voice mimicry are making distinctions between genuine consumers and software intermediaries increasingly subtle. Other technology developers, including the startup Instinct and Google with its Gemini assistant, have similarly developed autonomous voice-calling capabilities. Meta intends to transition Muse beyond mobile screens, preparing hardware expansions including intelligent eyewear and a dedicated wearable accessory termed the Muse Charm. Outbound calling constitutes one component of a broader operational portfolio, as Muse is also designed to send emails, assist with shopping, and coordinate travel reservations.