Google introduces live video avatars for Gemini enterprise agents
Google has launched Gemini 3.8 Live with Live Avatar, adding real-time visual personas to its voice agent models. These avatars can perform lip-syncing, natural expressions, and fluid turn-taking, and are designed for customer service and sales support. They are available only to Gemini Enterprise customers and include SynthID watermarking for transparency.
The Live Avatar feature builds directly on last week's Gemini 3.8 Live launch, which Google positioned as its most advanced real-time dialogue offering. The avatars combine visual presence with the existing voice models, enabling them to maintain conversation flow while accessing background data or executing tool calls mid-dialogue.
Organizations can choose from Google's preset avatar library or create custom personas from reference images, with enterprise allowlisting required. Every avatar carries SynthID watermarking to mark AI-generated content. Google demonstrated both realistic and stylized cartoon avatars, suggesting flexibility in how brands might deploy these digital representatives.
The introduction of lifelike AI avatars into customer service and sales roles could reshape how consumers experience everyday business interactions. People may find these digital representatives efficient and consistent, yet the "creepy" factor noted in the article suggests potential resistance. Trust could become a concern, as customers may struggle to distinguish human agents from AI avatars, despite SynthID watermarking. The technology's impact will likely depend on how naturally these avatars perform and whether businesses deploy them transparently.