Google is launching Guided Vision in Gemini Live across compatible Android devices, adding spoken descriptions of a camera view and verbal cues for framing it. In the company's announcement, the feature is presented as everyday visual assistance developed with blind and low-vision users.
The release combines two interactions: describing what a camera can see and helping the person holding it obtain a more useful view. That second part matters when a user cannot easily judge whether an item is centered, too close, or outside the frame. Google's stated safety boundary is equally important: the feature is not intended for navigation or obstacle detection.
A conversation about the camera view
Guided Vision adds conversational visual assistance to Gemini Live. After sharing the camera, a user can ask about an object or scene and continue with detailed follow-up questions. Google says the feature can proactively suggest reframing, such as panning slowly, tilting downward, or stepping back.
The announcement lists reading nutrition labels, appliance dials, washing-machine settings, and printed menus among its examples. It also describes searching nearby for a dropped earbud or an item in a spice cabinet, identifying colors and patterns, and receiving an overview of a room or objects on a table.
These are the company's intended uses, not a guarantee that every label, object, or scene will be interpreted correctly. A camera's position and the available visual context remain central to the interaction. The spoken reframing cues are designed to help obtain that context rather than leaving the user to guess what is missing from the frame.
Developed with accessibility testers
Google says it partnered with Aira on tens of thousands of hours of visual-interpretation data. More than 1,000 members of Aira's Trusted Tester network helped test and refine the model across daily routines, according to the announcement. Aira specialists also worked on establishing and evaluating safety guardrails.
The company describes feedback from blind and low-vision communities in India, Brazil, Singapore, Indonesia, Japan, and other locations. It says Guided Vision supports conversations in multiple languages. That work explains the development process, but the announcement does not provide an independently measured accuracy rate for a particular task or environment.
Google also identifies older adults, people with low literacy, and anyone trying to read fine print in poor lighting as potential users. The common interaction is spoken assistance with visible information, not a new medical or mobility service.
Where to enable it
Google says Guided Vision is available on Android 9 and above where Gemini Live's regions and languages are supported. It asks users to update the Gemini app and Android system software.
In the Gemini mobile app, the stated path is profile settings, then Use Guided Vision in Live. Users then open Gemini Live and share the camera. Android's Accessibility settings offer a Guided Vision shortcut under Vision assistance, with options including the floating accessibility button, a two-finger swipe, or both volume keys.
For TalkBack users, Google describes opening the TalkBack menu with a three-finger tap and selecting Guided Vision. These are the announcement's access paths; availability still depends on the supported device, region, and language.
Visual descriptions are not safe-travel guidance
Google explicitly says Guided Vision can make mistakes. It is not a medical device, mobility aid, or white-cane replacement, and it is not intended to provide safe-travel guidance or detect obstacles. Users should continue relying on established mobility aids and safe-travel practices.
That qualification should stay attached to descriptions of the product. Assistance with reading a package or exploring what is on a table is a different claim from safely guiding someone through an unfamiliar physical environment.