Google is giving its live AI agents a face. Gemini 3.8 Live now adds Live Avatar, a feature that combines near real-time video with live dialogue for customer service and sales support, creating a new visual layer for conversations with AI.
The feature arrived as Google launched Gemini 3.8 Live and Live Extended Thinking, which the company calls its most advanced live dialogue models. Live Avatar is aimed primarily at Google’s Gemini Enterprise customers, giving organizations a way to add animated visual representations to AI-powered interactions.
Gemini Live Moves From Voice Into Video
Live Avatar pairs near real-time visual presence with Gemini’s live dialogue models. Google describes the result as “an experience that listens, sees and speaks with a dynamic visual persona,” connecting conversation with an animated visual representation.
That representation can lip-sync during a conversation and display different facial expressions. Instead of relying on dialogue alone, Live Avatar adds movement and expression to the exchange, bringing video into the same live interaction as listening and speaking.
The feature focuses on customer service and sales support, where organizations can use visual AI models during conversations. Google’s Gemini Enterprise customers are the intended users, placing Live Avatar inside the business software space rather than presenting it as a general consumer feature.
Google announced Gemini 3.8 Live with Live Avatar on Sept. 24, 2026, at 7:59 PM UTC. The addition of avatars to Gemini 3.8 Live’s agents followed on Sept. 25, 2026, at 6:53 am EST.
One Avatar System, 97 Languages
Language support is a central part of the feature. Live Avatar supports 97 languages without degrading video fidelity or introducing visual drift, allowing the animated representation to keep its visual quality across supported conversations.
That combination connects spoken interaction, animated movement, and multilingual support inside one live system. The avatar is not limited to a single language experience, and Google says its video quality remains intact across all 97 supported languages.
Google will offer a library of preset avatars for organizations that want ready-made options. It will also allow organizations to create their own avatars, giving enterprise customers another path for shaping how their AI agents appear during customer service and sales conversations.
- Live Avatar adds video to Gemini 3.8 Live.
- The system can lip-sync and show different facial expressions.
- It supports 97 languages without degrading video fidelity or introducing visual drift.
- Organizations can choose preset avatars or create their own.
- The output includes Google’s SynthID watermark and safeguards designed to respect identity.
Visual AI With Identity Safeguards
Adding animated avatars to live conversations also brings questions about how those visual identities are created and used. Google says Live Avatar includes safeguards to “respect identity,” establishing a boundary around the avatars organizations create or select.
Output from Live Avatar carries Google’s invisible SynthID watermark. That watermark is part of the feature’s built-in approach to identifying generated output, while the identity safeguards address how visual representations should be used.
The combination of custom avatars, facial expressions, lip-sync, and live dialogue gives organizations more control over the presence of their AI agents. Preset options can support organizations that want a library choice, while custom creation gives them another way to shape that presence for business conversations.
Google’s launch also connects Live Avatar to the broader Gemini 3.8 Live platform rather than treating video as a separate experience. The same live dialogue models can now listen, see, speak, and present an animated visual persona during supported interactions.
For enterprise customers, that creates a new direction for AI customer service and sales support. The conversation is no longer limited to spoken responses, because the agent can now combine dialogue with a visual representation that lip-syncs, changes facial expressions, and supports 97 languages.
Google has positioned Gemini 3.8 Live and Live Extended Thinking as its most advanced live dialogue models, and Live Avatar pushes that launch into visual territory. With preset and custom avatars, multilingual support, SynthID watermarking, and safeguards designed to respect identity, Google is building a more visible form of enterprise AI.
The next phase of Gemini Live will now unfold through the avatars organizations choose and create. For Google’s Gemini Enterprise customers, live AI is gaining a presence that can listen, see, speak, and appear inside the conversation.
Based on
