Google has added a face to Gemini 3.8 Live, launching an enterprise feature that turns the company's live dialogue model into an animated, responsive avatar.
Google's announcement says Live Avatar pairs speech with near-real-time video generation. The result can lip-sync, shift facial expressions and take conversational turns while responding to what it sees and hears.
The feature is available starting Sept. 24 through Gemini Enterprise. It is not a general release in the consumer Gemini app.
Conversation continues while tools work
Google says Live Avatar can trigger tools and retrieve data in the background without stopping its dialogue. The company demonstrates that approach with a hotel check-in agent that keeps speaking while it handles separate tasks.
The system also supports speech-to-speech interaction across 97 languages, according to Google, adapting the avatar's lip movements and expressions when the conversation switches languages. Those are company-described capabilities rather than independent performance measurements.
Organizations can select from a library of preset characters. They can also create an avatar from a reference image while preserving a brand or character identity, but Google says custom-avatar creation currently requires enterprise allowlisting.
Watermarks for generated output
Google says all audio and video produced by Live Avatar is marked with SynthID, its imperceptible watermark for AI-generated media. The company also links a model card covering safety and deployment guidance.
Live Avatar extends Gemini 3.8 Live, which Google introduced the previous week as a native speech-to-speech model. This release adds a generated visual persona around that conversation rather than changing the underlying service into a consumer video-chat product.
