Since its launch, Gemini Live has been an impressive example of what conversational AI should sound like: natural, fluid and ready for interruptions. However, until now he suffered from one serious limitation: he was completely “blind”. Unlike OpenAI Advanced Voice Mode, Gemini Live relied solely on voice commands and text knowledge, with no awareness of what’s on your phone’s screen or in front of your camera.
Apparently, that will change soon. The Android Police team, digging through the latest beta version of the Google app (version 16.9.34.29), found clear evidence that Google is quietly preparing a massive, visual upgrade for Gemini Live.
What does “Project Astra” bring?
The hidden lines of code, first spotted by AssembleDebug, reveal a new camera icon that will appear within the Gemini Live interface. Text strings clearly describe the function: “Live with camera on”, “Camera pointed” and “Paused video”.
This directly corresponds with the announcements made by Google at the I/O conference last year, when they demonstrated “Project Astra”, a vision of an AI assistant that in real time “sees” and comments on the world around the user through the lens of the camera. Although the standard (text) Gemini was already able to analyze uploaded photos and the phone’s screen, the integration of this capability directly into Live mode represents a generational leap forward.
Imagine walking through a city abroad. Instead of taking a picture of the menu, translating it, and asking the AI for a recommendation, you’ll be able to simply point the camera at a restaurant, activate Gemini Live, and say, “See this menu? What local specialties do you recommend that’s not too spicy?” AI will analyze the image from the camera in real time and answer you by voice, without any intermediate steps.
Or, in a troubleshooting context: point the camera at the router and ask, “What do I need to do to reset it?”
When is this coming?
Although the code is already present in the application, the function is not yet active. History teaches us that Google often tests these features in the background for months before the official launch. Given that spring is traditionally the time for major Google announcements (like the upcoming Google I/O 2026), it’s very possible that we’ll see the “visual” Gemini Live in action in the coming months.
The race for the ultimate assistant
This upgrade is crucial for Google in the battle for dominance in the smart assistant market. While Apple Intelligence is still in the early stages of integration with Siri, and OpenAI is aggressively pushing its visual models, Google has the advantage of deep integration into the Android ecosystem. If they manage to make visual recognition in Gemini Live fast, accurate and energy efficient, it could redefine the way we use our smartphones every day, Android Police reports.