Google launches Gemini 3.8 Live with Live Avatar
Google has released Gemini 3.8 Live alongside a new capability called Live Avatar, extending its enterprise AI agent platform with the ability to present a speaking, animated video persona during interactions. Rather than a plain voice or text interface, agents built on this system can now appear as an on-screen figure that lip-syncs and responds in real time - a shift that could make AI-driven customer service or internal tooling feel considerably more familiar to end users.
The Live Avatar feature is aimed squarely at enterprise deployments, where companies build custom AI agents for tasks like support, onboarding, or sales assistance. Adding a visual persona to these agents doesn't change the underlying reasoning, but it does address a longstanding friction point: users often find voice-only or chat-only agents cold or hard to trust. A consistent, branded visual presence may help close that gap in professional settings.
On the technical side, Gemini 3.8 Live introduces background tool calls, which allow the model to query external systems or APIs during a conversation without interrupting the flow of dialogue. This is meaningful for agents that need to pull live data - inventory levels, appointment slots, customer records - without making the user wait through an obvious pause. Combined with speech support across 97 languages, the update positions the platform for global enterprise rollout without requiring separate localized builds.
Google's move into video avatars for AI agents puts it in a space already explored by a handful of startups, but integrating the capability directly into a broadly used enterprise AI platform at this scale is a different proposition. How well the avatar rendering holds up under real-world latency conditions, and how enterprises choose to customize the personas, will likely determine whether Live Avatar becomes a standard feature of AI-agent deployments or remains a niche option for specific high-touch use cases.
