01 What happened
Google announced Gemini 3.8 Live, integrating a new Live Avatar capability that allows AI models to maintain a visual presence during active dialogue.
02 Key details
- Live Avatar provides near real-time lip-syncing and expression adaptation supporting 97 languages.
- The system enables asynchronous tool execution, allowing the model to fetch data or trigger actions while continuing the conversation.
- A fully animated avatar can be generated from a high-quality reference image while preserving likeness and brand styling, but custom avatar creation is currently available only through enterprise allowlisting.
- All audio and video outputs from the system are embedded with SynthID watermarking for provenance tracking.
03 Why it matters
This update enables enterprises to deploy virtual agents that combine continuous visual engagement with the ability to perform complex background tasks.
04 Who it matters to
Software developers.
Original source