Meta rolled out real-time voice and video chat with Muse, its personal AI agent, to everyone on the app in the United States on Thursday, according to The Verge and TechCrunch.

The Verge reports that CEO Mark Zuckerberg announced the feature, while TechCrunch notes that Meta's chief AI officer, Alexandr Wang, posted a video on X demoing a conversation with his own Muse. Both outlets report that the avatar reacts with subtle expressions and gestures matching the tone of the conversation as it speaks.

A customizable avatar nicknamed Jolly

Both sources refer to "Jolly" as the nickname for Muse's default avatar, which users can swap for a different look, from a person to an animal, in the app's settings. TechCrunch describes Wang's avatar as a fluffy white dog that responded to a joke about his habit of posting memes on X by suggesting he add disclaimers to posts that could be misread as risque.

TechCrunch reports the feature runs on two new models, Muse Realtime Avatar and Muse Realtime Voice, which handle animating the avatar and generating speech in real time. The outlet adds that the voice and video calls also work on the web and on Mac, in addition to the mobile app.

From a limited test to nationwide access

The feature was first shown off at Meta's Connect developer conference in September, according to TechCrunch. Before this week, voice and video chat had been available only to a limited set of US users; Meta now says it is rolling out to the whole country, both outlets report.

How the conversation actually works

The Verge reports that users can change Muse's voice tone two ways: by asking the agent itself to adjust its tone mid-conversation, or by picking from preset options in the avatar settings. TechCrunch clarifies that the avatar doesn't use the device's camera to "see" the user — it only hears what's said and reacts based on that, with no access to video of the person talking.

TechCrunch notes Meta's reasoning for giving Muse a body and a name: the company says anthropomorphizing the agent this way makes users feel more connected to it, and that the connection should grow stronger as more people start having live conversations with the avatar.

In a hands-on test described by The Verge, Muse responded quickly to spoken questions, with occasional pauses while looking something up, the same behavior the outlet observed in text chat. The Verge describes the sync between speech and avatar animation as solid, though it found the result "a little uncanny" after setting its own avatar to look like Mark Zuckerberg.

Next stop: AI glasses and the Muse Charm wearable

The Verge and TechCrunch both confirm that Meta plans to bring the voice and video feature to its AI glasses "soon" and to Muse Charm, an unreleased wearable device with a small screen meant to display the avatar. Both outlets report that Meta expects to ship Muse Charm later this year.

In practice

  • Who can use it: everyone on the Muse app in the United States, per Meta; the feature also works on the web and on Mac, according to TechCrunch.
  • Since when: the nationwide rollout was announced on November 5, 2026; the feature was first demoed at the Connect conference in September.
  • What changed: voice and video chat with an animated avatar, previously limited to a small test group, is now available to the app's full US user base.
  • What's next: Meta says the feature is coming to its AI glasses and to Muse Charm, a wearable device expected to ship later this year.
  • Voice customization: users can ask Muse to adjust its own tone mid-conversation or pick from preset voice options in the avatar settings, per The Verge.