A visual presence in the conversation
According to the Last Week in AI channel, Gemini 3.8 Live Avatar adds a visible character to Google’s live AI conversations. Andrey Kurenkov and Jeremie Harris discuss the September 24 announcement in podcast episode 259. The feature combines speech with near-real-time video generation so that an avatar’s lips and expressions follow the conversation.
Google presents it for enterprise applications. Organizations can use preset characters or create a visual identity for their own service. The podcast considers how a face can change the experience of talking with an AI, alongside the model’s ability to process visual input.
Characters and customization
The official demonstrations show different avatar styles, including an illustrated animal character and a human-style avatar. A separate demonstration places the avatar beside a live camera view. These are examples of the interface Google is presenting for conversation, visual context and turn-taking.
Kurenkov and Harris connect this launch with the wider competition to make AI assistants feel more present in everyday interaction. The announcement centers on combining speech and a responsive visual persona, rather than replacing the organization’s task or service with an unrelated character.
