Aivora
AI Daily中文
Google DeepMindAI AgentIntermediate

Google DeepMind Introduces Gemini 3.8 Live with Live Avatar: Real-Time Visual AI Agents for Enterprise

Google DeepMind 發表 Gemini 3.8 Live with Live Avatar：具備即時視覺化身與背景任務處理能力的企業級 AI 代理

2 min read
中文English
Google DeepMind Introduces Gemini 3.8 Live with Live Avatar: Real-Time Visual AI Agents for Enterprise
The 30-second version

Building on Gemini 3.8 Live, Google DeepMind introduced Live Avatar to bring real-time visual personas to its dialogue models. It processes visual and audio inputs simultaneously to deliver natural expressions and precise lip-syncing. Crucially, it supports asynchronous tool calling, enabling the avatar to execute complex backend tasks (like checking in guests) in the background while maintaining an uninterrupted conversation. It supports seamless switching across 97 languages and integrates SynthID watermarks for safety.

Key points

01

Real-time Multimodal Presence

Pairs real-time video generation with speech, delivering precise lip-syncing and natural facial expressions for fluid conversations.

02

Asynchronous Tool Calling

Enables the avatar to trigger background tool calls and fetch data silently without disrupting the ongoing live dialogue.

03

97-Language Dynamic Adaptation

Dynamically adapts lip-sync and expressions across 97 languages mid-conversation without degrading video fidelity.

04

SynthID Watermarking

Integrates imperceptible SynthID watermarks directly into audio and video outputs to ensure AI-generated content is detectable.

How it works

Gemini 3.8 Live Avatar Architecture & Interaction Flow
Real-time inputTrigger background tasksReturn dataSync expressionsApply safetyPresent to userUser Input(Voice/Video)Live Audio/Video OutputAsync Tool ExecutionReal-time Avatar &Audio GenGemini 3.8 CoreReasoningSynthID Watermarking

Why it matters

Traditional virtual assistants suffer from pauses when executing tasks. By implementing asynchronous tool calling, Live Avatar can perform complex backend workflows (such as reservation checks) while maintaining an empathetic, uninterrupted conversation. This transforms AI agents from basic voice assistants into expressive, functional visual representatives for customer service and interactive guides.

Who it affects

  • Enterprise Leader
  • AI Developer
  • Product Manager

How to use it

  1. 1Smart reception & customer service: e.g., handling hotel check-ins in the background while maintaining polite live conversation with guests.
  2. 2Brand customized avatars: developers can generate responsive, expressive animated avatars reflecting their brand's identity using a single reference image.

Limitations & caveats

  • Custom avatar generation via reference images is currently restricted to enterprise allowlisted organizations.
  • Real-time, synchronized high-fidelity video generation requires robust bandwidth and may experience latency in poor network environments.

Related