On September 24, 2026, Google announced that Live Avatar, a new video avatar capability for its Gemini 3.8 Live conversational AI, is now generally available within the Gemini Enterprise platform. The feature adds a lip-synced video character that can hold a conversation while simultaneously carrying out background tasks such as searches or data entry. It is available across US and EU regions, and several industries, including automotive retail and call centers, have already put it into production.
A New Kind of Conversation That Keeps Working in the Background
Live Avatar is built on a native speech-to-speech foundation rather than a traditional pipeline that first converts speech to text before generating a reply. As a result, the assistant can keep track of context even when a user interrupts mid-sentence, and improved detection of speech boundaries has helped cut down response latency.
A key feature is that the avatar can continue talking with a user while simultaneously handling background work such as querying external systems or filling out forms. In an insurance intake scenario, for instance, the avatar can capture information during the conversation and automatically populate the relevant form fields in real time. The system can also process a live camera feed or screen share alongside audio, letting it respond based on whatever the user is currently looking at. It understands and speaks 97 languages, automatically detecting which one a user is speaking.
Custom Avatars and Provenance via SynthID
Businesses can choose from a set of ready-made avatar designs for immediate use, or, after completing an enterprise verification process, create a custom avatar from a single reference photo and an audio sample. Every piece of generated audio and video carries an invisible SynthID watermark, embedding provenance information that is not visible to viewers but confirms the content was AI-generated.
Early Adoption in Auto Retail and Call Centers
Autotrader, part of Cox Automotive, has deployed a shopping assistant built on Live Avatar. It searches and compares vehicles based on what a shopper says, highlighting relevant details on screen as it guides them through the process, replacing menu navigation with natural conversation.
Equal AI, which builds call center AI for the Indian market, uses Live Avatar to handle more than one million calls a day across nine Indian languages. The company highlights that its assistant goes beyond answering questions to actually complete tasks on a caller's behalf.
Salesforce has also said it is exploring how Live Avatar can be combined with its Agentforce platform, pairing real-time voice and video responses with agentic AI to improve customer interactions.
How Developers Can Access It
Developers can build custom assistants through the Live API, which supports real-time bidirectional streaming, and the Agent Development Kit, which handles agent definitions and session management. Businesses that need guaranteed performance can also opt into a provisioned throughput plan for dedicated capacity.
Summary
Google has added a video avatar capability called Live Avatar to its Gemini 3.8 Live conversational AI and made it generally available for enterprise customers. Beyond support for 97 languages, its standout features include the ability to keep working in the background while talking, custom avatar creation, and SynthID-based provenance for generated content. With early adoption already underway in auto retail and call centers, it marks another step toward automating customer-facing work with conversational AI.
