Ahead of I/O 2025, Google teases how it plans to take on ChatGPT with Gemini AI

Google Gemini AI

Google is doubling down on its assistant ambitions with Gemini, a new AI-driven experience that promises to be deeply personal, sharply proactive, and impressively powerful. Josh Woodward, VP at Google, laid out the vision in a detailed post on X, signaling a strategic shift toward an assistant that not only understands users but anticipates their needs—without constant prompting.

The first pillar of the strategy is personalisation. Dubbed “pcontext” internally, Gemini will soon begin using past chats and integrate context from across Google services—Gmail, Calendar, Search, Photos, and more—to deliver a more tailored experience, all with user permission.

Next comes proactivity. Gemini won’t just wait for instructions; it will surface relevant insights and suggest actions before you even think to ask. It’s an assistant that aims to reduce friction, allowing users to stay in flow.

The third leg of the plan, as per Woodward, is power. With models like Gemini 2.5 Pro and tools from Google DeepMind, the assistant will be able to research, generate content, and execute complex tasks—whether that’s building code, creating videos, or managing workflows.

Backing it all is Google’s vast infrastructure, including TPUs that enable these services to run at scale—and importantly, for free. Gemini is now available to all US students, with more countries to follow.

Woodward also highlighted a string of rapid-fire launches: Gemini 2.5 Flash, Veo 2 for video generation, LaTeX support, and a new image upload/editing feature. “Keep your feedback coming,” he wrote. “We try to listen, build, and iterate fast.”