Guided Vision: Real-Time Visual Assistance
Open Gemini Live, share your camera, and Gemini narrates your surroundings while telling you how to reframe: pan slowly to the right, tilt downward, step back. This is not a toy for sighted users but an accessibility feature built with the blind and low-vision community — trained in partnership with Aira on tens of thousands of hours of real visual interpretation, refined with more than 1,000 Aira testers. Live on Android 9 and above, in regions and languages where Gemini Live is supported
Tool Interface
Interactive tool will be available soon
Features
- ✓ Share your camera to get real-time spoken descriptions, ask follow-up questions and receive proactive spoken cues
- ✓ When the framing is off it prompts you to correct it: pan slowly right, tilt down, step back or move closer
- ✓ Typical uses: reading fine print and package labels, finding a dropped earbud, distinguishing colours and patterns, understanding an unfamiliar room's layout
- ✓ Trained in partnership with Aira on tens of thousands of hours of professional visual interpretation, stress-tested by more than 1,000 Aira trusted testers
- ✓ Aira specialists acted as subject-matter experts to help establish and evaluate the safety guardrails
- ✓ Available on Android 9 and above in regions and languages where Gemini Live is supported, with the Gemini app and system software updated
How to Use
- Check the device first: Android 9 or above, in a region and language where Gemini Live is supported
- Update the Gemini app and Android system software to the latest versions, or Guided Vision may not appear
- Start a Gemini Live session, share your camera and turn Guided Vision on
- Ask in plain language — what colour is this, what does the label say — and ask it to repeat or reframe if audio is unclear
- Follow the spoken reframing cues, let it get a clear view before asking for detail; answer quality changes noticeably
FAQ
What is Guided Vision?
A real-time visual assistance capability in Gemini Live: share your camera and Gemini describes what is in front of you in speech while offering natural reframing cues. Google's announcement: https://blog.google/innovation-and-ai/products/gemini-app/guided-vision-gemini-live/
How is it different from ordinary multimodal Q&A?
The difference is proactive guidance and accessibility-first design. It speaks up when your framing is off instead of waiting to be asked, and the whole capability was developed with the blind and low-vision community.
Which devices support it?
Android devices running Android 9 and above, in regions and languages where Gemini Live is supported. The Gemini app and Android system software need to be up to date.
Where does the training data come from?
Google partnered with Aira on tens of thousands of hours of professional visual interpretation, with more than 1,000 Aira trusted testers stress-testing and refining it, and Aira specialists helping establish and evaluate safety guardrails.
What can it be used for?
The official examples include reading nutrition labels on food packaging, deciphering appliance dials and washing machine cycles, reading printed menus in dim restaurants, finding a dropped earbud or the black pepper in a crowded spice cabinet, checking whether a striped shirt pairs with trousers, and getting an overview of an unfamiliar room.