Google has announced a major expansion of AI capabilities in Search at I/O 2026, introducing a redesigned AI-powered Search experience, intelligent Search agents, generative UI tools, and expanded Personal Intelligence features powered by Gemini 3.5 Flash.
According to the company, AI Mode has now surpassed one billion monthly users globally. Google also says AI Mode queries have more than doubled every quarter since launch, while overall Search queries reached an all-time high last quarter as users increasingly relied on conversational and AI-assisted searches.
Gemini 3.5 Flash becomes default AI Mode model
Google is now upgrading AI Mode with Gemini 3.5 Flash as the default model globally. The company says the latest Flash model is designed to deliver sustained frontier performance for agentic workflows, reasoning, and coding tasks directly inside Search.
Google has also introduced what it describes as the biggest Search box upgrade in more than 25 years. The new AI-powered Search box is designed to support more natural and conversational prompts instead of traditional keyword-based searches.
The Search field can dynamically expand for longer queries and provide AI-generated suggestions that go beyond standard autocomplete to help users better formulate requests.
Search now supports multimodal inputs including:
- Text
- Images
- Files
- Videos
- Chrome tabs
Google says users will continue receiving a range of Search results similar to the existing Search experience.
Conversational Search improvements
Google is also expanding conversational capabilities in Search across desktop and mobile platforms.
Users can continue asking follow-up questions directly from AI Overviews and move into AI Mode without restarting a Search session. According to Google, conversational context remains active during follow-up interactions, while links and supporting articles become more relevant as discussions continue.
Search agents
Google is introducing Search agents that allow users to create, customize, and manage multiple AI agents directly within Search.
The first rollout focuses on information agents that operate continuously in the background. These agents can intelligently reason across information sources and monitor blogs, news websites, social platforms, shopping data, finance updates, sports information, and other real-time sources.
Google says the agents can provide synthesized updates and take actions based on user-defined requests.
Examples shared by the company include:
- Monitoring apartment listings matching specific requirements
- Tracking sneaker drops and athlete collaborations
- Sending real-time notifications related to ongoing topics
Google is also expanding agentic booking capabilities in Search for local experiences and services. Users can provide detailed requirements, while Search can surface pricing, availability information, and direct booking links.
For selected categories including:
- Home repair
- Beauty services
- Pet care
Google can also contact businesses on behalf of users.
The company additionally announced new agentic shopping capabilities in Search, with separate details shared through its Shopping announcements.
Agentic coding and generative UI in Search
Google is bringing Google Antigravity technology and Gemini 3.5 Flash agentic coding capabilities directly into Search.
According to the company, Search can generate responses in different formats dynamically and build custom generative UI experiences in real time based on user queries.
These generated experiences can include:
- Interactive visuals
- Tables
- Graphs
- Simulations
- Custom layouts
Google says the system can generate visual tools for topics such as astrophysics or mechanical systems.
The company is also introducing AI-generated dashboards, trackers, and mini apps for recurring tasks including:
- Wedding planning
- Home moving
- Fitness tracking
Users can return to these mini apps over time to continue tracking progress and ongoing activities.
Google says these experiences can use live information from maps, reviews, weather data, and other real-time sources.
Personal Intelligence expansion
Google is expanding Personal Intelligence support in AI Mode to nearly 200 countries and territories across 98 languages without requiring a subscription.
Users can securely connect services including:
- Gmail
- Google Photos
- Google Calendar support coming later
Google says Personal Intelligence was designed around transparency, choice, and user control, allowing users to decide when app integrations are enabled or disabled.
Availability
- Gemini 3.5 Flash is now the default model in AI Mode globally
- The new AI-powered Search box and conversational AI Overview improvements are rolling out globally in supported AI Mode regions and languages
- Generative UI capabilities in Search will roll out globally this summer free of charge
- Information agents and Antigravity-powered mini apps will launch first for Google AI Pro and AI Ultra subscribers this summer
- Agentic booking capabilities for local services and experiences will begin rolling out in the U.S. this summer
- Personal Intelligence support is expanding to nearly 200 countries and territories across 98 languages without requiring a subscription