Quick answer
September 24, 2026: Google announced Gemini 3.8 Live Avatar, a real-time animated presenter driven by Gemini that speaks and lip-syncs in 97 languages, for use in Meet, customer-facing products, and Workspace; and began testing "Call for Me" on Pixel 11, where Gemini phones a business, handles the conversation, and reports back. Live Avatar is rolling out to Workspace and developer customers; Call for Me is a limited US test.
Two announcements, one theme: Google is putting a face and a voice on Gemini and letting it act in the world on your behalf. Both features have obvious uses and obvious questions.
Live Avatar
- A real-time avatar that speaks Gemini's responses with synchronised lip movement in 97 languages
- Intended for customer service, training, Meet presentations, and multilingual communication
- Available to Workspace and Google Cloud developers, with stock avatars and consent-gated custom ones
- Competes with Synthesia, HeyGen, and Tavus, but as a platform capability rather than a standalone product
Call for Me
Tell Gemini "find out if the pharmacy has my prescription" and it calls, speaks with whoever answers, and summarises the result. Google says the call identifies itself as an AI assistant. It builds on the Duplex work Google demonstrated years ago and on Gemini's voice mode, and it is being tested on Pixel 11 in the US before any wider rollout.
The questions
- Disclosure: businesses receiving AI calls need to know; Google's stated approach is that the assistant identifies itself
- Consent for custom avatars: Google gates cloning a real person's face behind verification
- Accuracy: a summary of a phone call is only as good as the transcription and understanding; expect errors
- Regulation: several jurisdictions are drafting rules on synthetic callers and presenters
Bottom line
Live Avatar is a direct threat to standalone avatar tools; Call for Me is a preview of assistants that act by phone. Both are early, both are Google-only for now, and both will be copied.
