Meta gives its Muse AI agent a live video avatar

At Meta Connect 2026 in Menlo Park, Meta detailed an expansion of Muse, its AI agent. TechCrunch reported that Muse users will receive a digital avatar with a body, face and voice — referred to by Mark Zuckerberg as "Jolly" — and will be able to "have live conversations with this avatar, giving it tasks and asking it questions." A new AI model, Muse Realtime Avatar, has been developed to power the feature. The article gives no concrete launch timeline, noting only that the capability is coming "soon."

Other Muse announcements at the event included integration with Meta smart glasses via a wake word, Mac automation, a dedicated email address for the agent, retail partnerships with Shopify, Best Buy, Sephora, Walmart, Wayfair, Expedia and Instacart, payment integrations with Stripe, Shop Pay and PayPal, and a developer connector platform that received more than 1,500 applications in one week. Zuckerberg said: "The centerpiece of our vision for what we're building is Muse. In the coming years, I expect that Muse is going to grow into the personal superintelligence..."

Why it matters for digital humans: Until now, giving a general-purpose assistant a face has been the business of specialist vendors. A platform holder shipping a named, real-time avatar model as the default front end for its consumer agent changes the baseline expectation for what a conversational interface looks like, and it puts an in-house model directly alongside the third-party avatar stacks teams currently license. Worth watching two things the announcement does not settle: latency and turn-taking quality under real network conditions, and whether Muse Realtime Avatar is exposed to developers or stays locked to Meta's own surfaces. No availability date, pricing or performance figures were given, and nothing shown has been independently benchmarked.