Meta’s consumer AI agent Muse can now call U.S. businesses for you, but only if you pay for it. TechCrunch AI reports that the feature first went into testing in September and has now rolled out to every paid subscriber. Before this, only a few early adopters who asked Muse for access could use it.
Meta says Muse passed 10 million users last month. Calling is part of its push to make Muse a mainstream assistant that clears real chores off your to-do list instead of just answering questions.
📞 What Muse Can Do Now
You ask Muse to call a business, and it makes the call for you. Alexandr Wang, Meta’s chief AI officer and head of its Superintelligence Labs division, gave a few examples of calls people would happily hand off:
- Your cable company, along with the phone trees and hold music that come with it
- The DMV, a call nobody looks forward to
- Restaurants without online booking, where you’d otherwise have to call to get a table
💰 What It Costs
Calling is only available on paid plans. According to TechCrunch AI, Muse has two tiers:
- $20 per month for 500 million Muse tokens per week. This is the entry plan, and it’s enough for most people who want an assistant to run occasional errands.
- $100 per month for 3 billion tokens. This one is for heavy users who lean on Muse throughout the day.
There’s also something for free users. Wang said this week that Muse email addresses, which launched recently, will soon be available on the free tier. So Meta is putting some agent features in front of everyone while keeping the more expensive ones, like calling, behind the paywall.
🔍 Five Things Worth Knowing
- Muse isn’t first. The startup Instinct already offers AI calling to businesses, and so does Google’s Gemini. Meta’s edge here is reach, not novelty, since few companies can put a feature in front of as many people.
- Businesses often hang up. TechCrunch AI notes that many businesses and receptionists end the call as soon as they realize they’re talking to an AI. The technology works fine. Getting the person on the other end to stay on the line is the hard part.
- That’s going to get harder to spot. AI voices sound more human every year, so telling a real customer from an AI calling on their behalf will get tougher. That raises fair questions about disclosure, and about whether businesses will start screening these calls automatically.
- Bad customer service makes the pitch easy. People have spent years stuck in phone trees, waiting on hold, and getting transferred from one department to another. An AI with unlimited patience sounds great to anyone who’s lost an afternoon to an airline helpline.
- Pricing is part of the strategy. Putting calls behind a $20 paywall gives Meta a clear reason for people to upgrade. Each call also costs real money to run, so it makes sense that Meta isn’t giving it away yet.
🧭 Why It Matters
Calling is one of the clearest tests of an AI agent. A chatbot gives you information. An agent goes out and does something for you, in a setting it doesn’t control, with a stranger who may or may not want to cooperate. If Muse can reliably book a table or sort out a billing problem, it’s doing a job you’d otherwise do yourself.
I think the bigger story is what happens on the other side of the phone. Once millions of consumers send AI agents to make their calls, businesses will need their own AI to answer them. That could leave us with AI talking to AI on routine requests, and the old customer service phone line might quietly turn into a channel between machines.
⏭️ What Comes Next
Things to watch: whether Meta expands calling outside the U.S., whether it eventually comes to free users, and how businesses respond. More businesses picking up for AI callers would make this feature much more useful. More businesses hanging up would show its limits. You can find the full details in TechCrunch AI’s original report.