Google used its I/O 2026 developer conference to roll out a set of updates to the Gemini app that push its AI beyond simple question-and-answer tasks. The new version introduces proactive agents — AI that can start actions on its own without waiting for a user to type a prompt. On-demand media generation and deeper task automation across Google services are also part of the package.
Agents that act first
The headline feature is a shift from reactive to proactive assistance. Instead of a user having to ask for help, Gemini's agents can now initiate tasks based on context, habits, or inferred needs. Google didn't detail every trigger, but the company described the agents as capable of acting without explicit user prompts — a step beyond the chatbots and voice assistants that require a command before they do anything.
Built-in media creation
Media generation is now baked into the Gemini app. That means the app can produce images, audio, or short video content on demand, all from within the same interface. The company didn't specify file formats or resolution limits, but the feature is designed for quick, on-the-fly creation rather than professional production work.
Automation across Google's tools
The updates also tie Gemini more tightly into Google's ecosystem. Task automation flows across services like Gmail, Calendar, Drive, and Docs, letting the AI string together actions without the user jumping between tabs. For example, an agent could pull an attachment from an email, summarize it in Drive, and drop a reminder into Calendar — all triggered by a single user request or even an automatic cue.
A strategic pivot for Google AI
These updates mark a deliberate turn in Google's AI strategy. The company is moving away from purely reactive tools — ones that answer only when spoken to — toward what it calls proactive, user-centric assistance. The Gemini app is now the hub for that vision. Whether that means fewer manual tasks or more autonomy for the AI, developers at I/O got a glimpse of a system designed to anticipate needs rather than wait for commands.
Google hasn't announced a public rollout date for the new Gemini features. Attendees at the conference could test demos, but the company offered no timeline for when the proactive agents and media generation would land in users' hands.
