Google's Gemini AI assistant can now tap into users' search histories to provide more personalized responses, marking a significant expansion of the chatbot's capabilities.
The new feature requires explicit user permission and runs on Google's experimental Gemini 2.0 Flash Thinking model, which launched in late 2024 as the company's first reasoning-focused AI system.
Users who enable personalization in Gemini apps allow the AI to analyze their search patterns and incorporate this information into its responses. Google emphasizes transparency - users can turn off the feature anytime, and Gemini clearly indicates which data sources inform its personalized answers.
For testing the new capabilities, Google recommends asking Gemini about vacation spots, YouTube content ideas, or potential new hobbies. The system then draws on individual search histories to make tailored suggestions. Users can review and manage their search data through Google's activity controls page.
Expanding across Google's ecosystem
The company plans to extend personalization beyond search in the coming months, integrating data from Google Photos and YouTube. This expansion aims to give Gemini a more comprehensive understanding of user preferences and activities.
Currently, the feature is available to Gemini and Gemini Advanced subscribers through the web interface, with mobile access coming soon. While the system supports more than 45 languages and operates in most countries worldwide, it notably excludes the EU, Switzerland, and the UK.
This search-based personalization represents a step toward Google's vision of a universal AI assistant that works across all platforms, modes, and devices. The company's extensive user data collection gives it a unique advantage in this space. The complete assistant ecosystem includes app control through Assistant, website navigation via browser agent Mariner, and real-world interaction through Astra's audiovisual capabilities.
