At Google I/O 2025, the focus was on extending Gemini to create a universal AI assistant. This involves evolving Gemini into a “world model,” capable of planning and simulating experiences, similar to how the human brain functions. This builds upon advancements in agentic capabilities, such as Project Mariner which allows for multi-tasking within the browser. The goal is to integrate these features into products like Gemini Live, Search, and new form factors like glasses, with a strong emphasis on safety and responsible development.