Point your camera at anything broken. Talk naturally. FixIt Genie is a multimodal live agent that sees what you see, hears what you say, and walks you through the fix — in real time.
Bidirectional audio streaming means you talk naturally, interrupt naturally, and get real-time multimodal guidance without push-to-talk friction.
A quick walkthrough of the product flow: clear onboarding, a focused start state, and a live transcript that stays readable while the agent listens and talks back.
Three live streams — camera frames, voice in, voice out — multiplexed over a single WebSocket to a Google ADK agent on Cloud Run.
There are 15 million skilled trade workers in the US alone. Every day they face equipment they've never seen before, error codes with no obvious cause. The standard solution — manuals, tutorials, hotlines — all have the same flaw: they can't see what you see. Gemini Live changes that constraint.
Read the full post →