Meta's Muse Can Make Phone Calls, but Access Is Still Limited
Muse, the AI agent from Meta, can place phone calls to U.S. businesses on a user's behalf, but the feature is in limited early access and most accounts cannot use it yet. That is the main finding of an October 2026 analysis from voice AI company Bland, which argues the tool is no substitute for business calling infrastructure.
- Muse can call U.S. businesses on a user's behalf, but the feature is in limited early access that began around September 2026.
- Targeted tasks include restaurant bookings, business hours checks, appointment changes and delivery confirmations, according to nullbot newsroom.
- Bland claims Meta used human concierge workers to complete calls and did not disclose this to users.
- Bland says Muse handles one call at a time and is not built for high-volume business calling.
- The analysis comes from Bland, a vendor of AI phone calling infrastructure.
What the calling feature does
According to Bland's post, which cites coverage by nullbot newsroom, the feature began rolling out incrementally around September 2026. Users who had already requested early access were prioritized. Bland describes this as a deliberate approach, with Meta testing the feature in controlled conditions before widening availability. Anyone who asked for access and has not received it is having the expected experience, not hitting a technical error, the post says.
The scope is narrow. The feature covers calls to U.S. businesses made for the user. International businesses and personal contacts fall outside it. According to nullbot newsroom, the targeted tasks include booking a restaurant, checking business hours, moving an appointment, and confirming a delivery window.
Bland says Muse places an outbound call, listens for a human or automated greeting, and works through phone menus by choosing the right branch. It will also wait on hold, which Bland calls the feature most users notice first.
Where the limits sit
Bland lists several boundaries. Muse will not make emergency calls, medical advisory calls, or calls where the outcome must be guaranteed, according to the post. It also says Muse declines to proceed when a call requires judgment or negotiation.
The post also makes a claim about how the feature works. It says Muse's calling was never fully autonomous and that Meta embedded human concierge workers to finish calls the AI could not complete, partly because businesses were hanging up on AI voices. Bland states that this human fallback was not disclosed to users. The article does not name a primary source for that claim, so it should be read as Bland's account.
Bland also says Muse handles one call at a time. If a hold queue runs long and a second call is needed, the second task waits.
What it means for businesses
Bland's central argument is that a consumer agent making occasional errand calls belongs to a different category from the voice infrastructure a customer experience or operations team needs. Such teams, the company says, need to run many concurrent calls, keep an audit trail, and control how data moves during a conversation. In its view, treating the two as the same leads to premature pilots and wasted effort.
The source is a vendor blog. Bland sells AI phone calling infrastructure and presents its own product as the answer to the gaps it describes, including dedicated numbers, self-hosted voice agents, and integration with contact center platforms such as Amazon Connect. Its claims about Muse's limits and about its own platform come from the company itself.
For individual users, the practical point is availability: most people asking whether Muse can make calls cannot yet try it. For teams considering AI calling for business use, the post advises checking concurrency, compliance and auditability before building a workflow around any consumer tool.
Bland AI: Can Muse Make Phone Calls? Now It Can With Its Own Number
