StreetReaderAI is making street-level navigation more accessible for blind and low-vision users. Our new prototype uses context-aware, real-time Gemini models to convert Street View imagery into a dynamic, conversational guide opening up new possibilities for exploration.