Google has developed Gemini Intelligence, an advanced AI agent aimed at improving user interaction with mobile applications by managing multi-step tasks. It can interpret information from emails, such as analyzing a class syllabus to compile a shopping cart for textbooks. Gemini can also utilize contextual information from a user's phone screen or images to assist in real-world scenarios, like finding tours based on travel brochures.
To address user privacy concerns, Gemini will only initiate tasks when explicitly instructed, require user confirmation for purchases, and allow users to manage data access through a permissions menu. A progress bar feature enables users to stop the agent's activities at any time. Gemini Intelligence is set to launch on the latest Pixel and Samsung Galaxy phones, with its success depending on reliability and user experience compared to other AI agents.