Google has introduced Live Avatar alongside its Gemini 3.8 Live update, enabling users to interact visually with real-time animated AI personas. Currently restricted to Gemini Enterprise customers, the system supports lip-syncing and fluid facial expressions across 97 languages without losing visual fidelity or introducing drift during conversations.

The feature builds on Gemini 3.8 Live’s ability to process visual inputs in near real-time and present on-screen information dynamically. Google allows enterprise clients to select from preset avatars or create custom models, embedding SynthID invisible watermarks and identity safety guardrails into outputs.

Why it matters

  • Provides enterprises with real-time, multilingual visual interface options for customer-facing and internal AI agents.

  • Demonstrates advances in low-latency multimodal generation and synthetic video synchronization.

Source: theverge.com