Google Unveils Gemini 3.8 Live with Real-Time AI Avatars for Enterprises

Google has officially launched Gemini 3.8 Live, introducing a groundbreaking feature called Live Avatar that provides near real-time animated AI avatars for enterprise applications. This innovative addition allows avatars to lip-sync and express emotions during conversations, significantly enhancing the visual and conversational capabilities of AI.

The Live Avatar feature is designed for various applications, including customer service and interactive walkthroughs. It enables the AI to process both audio and visual inputs simultaneously, allowing agents to respond to user queries while interpreting live camera feeds or screen shares. Furthermore, the system supports asynchronous tool calls, enabling agents to query APIs or backend systems without interrupting the conversation.

Google's Live Avatar can automatically detect and switch between 97 languages, adapting speech and expressions without requiring manual adjustments from users. Companies can choose from a library of preset avatars or create customized versions based on reference images, although custom avatars are currently available only to approved enterprise customers.

To ensure the authenticity of AI-generated content, Google is implementing SynthID watermarks on both audio and video outputs. Gemini 3.8 Live with Live Avatar is now available through Gemini Enterprise and the Gemini Live API, with endpoints in the US and EU. Meanwhile, the Gemini 3.8 Live Extended Thinking feature remains in private preview.