At I/O 2025, Google showcased its goal of turning Gemini 2.5 Pro into a universal AI assistant. The company aims to evolve it into a “world model” capable of planning, imagining new experiences, understanding complex scenarios, and taking actions—much like how the human brain operates.
Since Gemini 2.5 Pro is natively multimodal, Google says it can serve as the foundation for a universal AI assistant within the Gemini app that can multitask on your behalf. The goal is to expand Project Astra also known as Gemini Live by making it more action-oriented and capable of handling complex tasks seamlessly.
Right now, Gemini Live lets you share your screen and camera to interact with the AI in real time. But Google is testing smart actions as part of the experience. Soon, Gemini Live could browse the web, book appointments, make purchases, conduct research, and more all while you stay connected.
For instance, you can ask Gemini Live to open YouTube and play a specific video. It can also download user manuals, scroll through them, and find the exact information you need. When using the camera, Gemini can highlight specific parts directly on your screen.
Google says these features will be built into the Gemini app, turning it into a true universal AI assistant. It will understand context, plan your request, and take action on your behalf. The early prototype of Project Astra is currently being tested and will be integrated into the Gemini app down the line.
