What happened
Google has updated its Gemini 3.8 Live with Live Avatar service to allow enterprise customers to generate custom, animated avatars based on a single reference image. This , which is currently restricted to enterprise allowlisting, enables organizations to create branded visual personas that maintain specific likenesses and styles while utilizing the model's existing real-time, multilingual, and asynchronous tool-calling capabilities.
Google's update to the Gemini 3.8 Live with Live Avatar platform introduces the ability for organizations to upload a single reference image to generate a fully animated, responsive avatar. This capability is designed to preserve the specific likeness, brand styling, or character identity of the source image.
The is currently available exclusively through an enterprise allowlisting process. This controlled rollout suggests that Google is managing the deployment of custom visual identities to ensure quality and safety standards are met before broader availability.
The custom avatars retain the core functionality of the standard Live Avatar service, including near real-time lip-syncing, natural facial expressions, and the ability to switch seamlessly between 97 languages. The system also supports asynchronous tool calling, allowing the avatar to perform background tasks without interrupting the visual flow of the conversation.
Why it matters
The introduction of custom avatar generation marks a significant shift in how enterprises can deploy conversational AI. By moving beyond generic, preset characters, businesses can now align AI agents with their specific brand identities. The ability to maintain visual consistency while leveraging Gemini’s advanced reasoning and multilingual lip-syncing capabilities allows for more personalized and professional customer-facing interactions, potentially increasing user engagement in sectors like hospitality, retail, and digital support.
For enterprises, the ability to customize the visual persona of an is a critical step toward integrating AI into customer-facing workflows. It allows for a consistent brand experience that is not limited by a library of generic presets.
The integration of SynthID into both the audio and video output of these custom avatars is a notable safety measure. As businesses increasingly rely on AI for communication, the ability to detect and verify AI-generated content is essential for maintaining trust and preventing misinformation.
The combination of visual presence with Gemini's underlying reasoning capabilities allows for more complex, task-oriented interactions, such as hotel check-ins or service walkthroughs, which were previously difficult to manage with standard text-based or voice-only AI interfaces.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
The primary focus for enterprise adoption will be the performance of the custom avatar generation under the current allowlisting process. Observers should monitor how effectively these custom avatars maintain visual fidelity during complex, long-form interactions and how the integration of SynthID impacts the perceived authenticity of these branded personas in real-world, high-stakes customer service environments.
The effectiveness of the allowlisting process in managing the creation of custom avatars will be a key indicator of how Google intends to scale this technology while mitigating risks related to identity impersonation.
Future updates may clarify the technical requirements for the reference images and the latency impact of generating custom avatars compared to the pre-built library options.
Industry adoption rates will likely depend on how well these custom avatars handle edge cases in multilingual environments and the reliability of the background tool-calling features during high-volume enterprise use.