Google rolls out Guided Vision inside Gemini Live, adding real-time voice descriptions from the phone camera

Google activated Guided Vision inside Gemini Live on compatible Android devices yesterday. The feature shares the camera feed with the model and returns a live voice description of whatever the lens captures — small print, surrounding objects, details on a specific item. The launch arrived as part of recent Pixel updates.

In practice, a user can ask Gemini to read an expiration date on a recognized product, identify a lost object, or explain what is written on a remote control. If the requested object is not in frame, the model speaks directions on how to point the phone. Follow-up questions about what is visible continue the conversation without restarting the session.

The tool was designed primarily for people who are blind or have low vision, as well as anyone who needs occasional assistance in specific situations. The concept parallels Apple's VoiceOver Live Recognition on iPhone and Vision Pro — real-time scene recognition and narration — except Google's implementation lives inside Gemini Live rather than as a separate system feature.

Beyond the Gemini app, Guided Vision is also accessible through Google TalkBack, and users who prefer quick access can set an accessibility shortcut in system settings. The minimum requirement is Android 9, so the range of supported devices is relatively broad.

Google explicitly states that the feature is not a substitute for a white cane, is not intended for navigation, safe driving guidance, or obstacle detection. The warning appears in the launch documentation to prevent over-reliance on a voice description that is not guaranteed to be accurate in dynamic environments.