Conversations that do more.
Speech-to-speech support, live product guidance and multimodal interfaces. Connect answers to your knowledge and the next action.
AI, put to work.
New intelligence.
Real possibilities.
We build AI into the way people speak, work and interact with the world.
Try it for yourselfSpeak. Interrupt. Keep going.
A natural conversation with a native audio model.
Try live voiceGive it a job. Watch it work.
An agent checks orders, reads stock and prepares the next step.
Run a workflowLet intelligence see your world.
Explore how a model reads objects and relationships in a space.
Explore a sceneFrom a model
to a working system.
Connect intelligence to the context, tools and people that make it useful.
Work that moves itself forward.
Agents that coordinate tasks across your systems, with clear permissions, review points and recovery when something changes.
Intelligence beyond the screen.
Vision, sensors and embodied reasoning. Explore Gemini Robotics and physical AI around a specific environment, task and hardware.
A more human interface.
Live avatars and shared visual context can make guidance feel more immediate. We evaluate the interface against the job it needs to do.
Built for change.
We evaluate new models against real tasks. The model can change; the purpose should stay clear.
Technology notes · reviewed 16 October 2026
Gemini 4 Argon, released 13 October, powers the workflow trial. Voice uses Gemini 3.8 Live. Each trial is a TTL experiment; it is not a past client delivery.
Robotics and avatar integrations depend on the model, hardware and access available for a project. The scene trial demonstrates image reasoning, not robot control.
Google’s release notes ↗