Google's new Gemini Live Avatars want to make support bots feel more human

Last week, Google introduced us to new members of the Gemini 3.8 family, which included Live and Live Extended Thinking. Continuing on the Gemini 3.8 Live train, the company is rolling out something new. While it won’t be a product for consumers, you may see it the next time you talk to an agent bot.

If you hate talking to AI support bots, maybe you’ll change your tune after they get a Live Avatar update. Google has announced that it is rolling out Gemini 3.8 Live with Live Avatar, a visual persona that combines real-time video generation with speech models. According to the tech giant, these avatars are designed to provide “engaging customer service” or “interactive walkthroughs” with the help of their “precise lip-syncing, natural expressions, and fluid turn-taking.”

Essentially, Live Avatars are meant to give an enterprise agent more personality. Google says this will generate “enriching conversations for a more comprehensive experience.” Adding to that, Live Avatar can adapt its lip-syncing and expressions as it switches between languages. There’s support for 97 languages in total, and switching languages has no negative effect on the video quality.

Outside of their visual presence, Live Avatars are powered by Gemini’s advanced reasoning and asynchronous tool calling. This allows the avatar to trigger tool calls and fetch data in the background without interrupting the conversation. For example, you could ask the avatar to check you into a hotel as you talk about other matters.

Gemini 3.8 Live with Live Avatar is launching today. However, it’s only available for Gemini Enterprise subscriptions.