Gemini 3.8 Live Avatar Introduced: Google introduced Gemini 3.8 Live with Live Avatar on September 24, 2026. The technology combines real-time speech with animated video avatars, bringing visual presence to AI conversations. Available through Gemini Enterprise, it targets customer service, interactive assistance and other business applications.
AI Avatars Show Human-Like Expressions: Gemini Live Avatar synchronises facial expressions and lip movements with spoken responses. Its video generation operates at 24 frames per second, creating animated digital characters. These capabilities allow businesses to develop conversational agents that communicate through both speech and visual expressions.
Conversations Become More Natural: Gemini 3.8 Live supports fluid conversations, including interruptions and changes in speaking turns. Its native speech-to-speech architecture helps maintain context during interactions. The system also processes audio and visual information together, enabling more context-aware responses during customer support and assistance.
AI Understands Real-Time Visual Context: The technology can process live camera feeds and screen sharing alongside spoken instructions. This allows AI agents to respond to information users show them, rather than relying exclusively on text or voice. Potential applications include troubleshooting, guided assistance and interactive customer support.
AI Conversations Cross Language Barriers: Gemini 3.8 Live supports 97 languages and automatic language detection. Its multilingual speech and video synchronisation allows avatars to adapt their expressions and lip movements during language changes. This capability could help businesses serve customers across different regions through conversational AI interfaces.
AI Avatars Can Complete Tasks: Gemini Live Avatar can execute API calls and use tools in the background while continuing conversations. For example, a hotel agent could process guest check-ins without pausing the interaction. Businesses can use preset avatars or develop custom characters through enterprise-approved access.
Security and Transparency Remain Important: Google embeds SynthID watermarks into generated audio and video to help identify AI-created content. Custom avatars also require enterprise approval and verification. These safeguards address identity misuse and transparency concerns as businesses introduce increasingly human-like AI agents into customer-facing services.