Google expands Guided Vision as smartphones become accessibility assistants

Google's September Android feature cycle includes Guided Vision for Gemini Live, designed to provide spoken descriptions of what the phone camera sees and actively help users frame the camera correctly. Instead of merely analysing a static photograph, the system can tell a user to pan, centre an object or reposition the camera while trying to read fine print, identify household items or interpret a poorly lit menu.

Guided Vision is designed to be accessible through Android's accessibility shortcuts and TalkBack, with support planned across a broad range of devices where Gemini is available. This is one of the strongest examples of generative AI solving a concrete smartphone problem rather than functioning as a novelty chatbot. Real-time visual assistance can materially increase independence for users with low vision, provided recognition is reliable.

The feature also illustrates why mobile AI increasingly depends on tight integration between camera hardware, operating-system accessibility services and multimodal models rather than a standalone application.