The search bar is no longer a place to type queries. It is now a place to assign work.
Google unveiled its most radical search overhaul in nearly 30 years at I/O 2026 on Tuesday, transforming the iconic text box into an expandable, multimodal command center powered by the new Gemini 3.5 Flash model. The search field now grows to accommodate long, conversational queries and accepts text, images, videos, files, and Chrome tabs as input. But the bigger story is what happens after you hit enter.
Google introduced Search agents that operate 24/7 in the background, monitoring topics and sending updates without requiring users to repeatedly check for new information. Track stock movements based on specific conditions, monitor apartment listings, or get notified when a favorite athlete announces a sneaker collaboration.
CNN reported users can enter a query like "Keep me updated when any of my favorite athletes announce sneaker collabs or signature drops" and the agent handles the rest.
These information agents launch this summer for AI Pro and Ultra subscribers in the U.S. first. The search box now generates custom mini apps and dashboards in real time. Ask about astrophysics and Search can produce an interactive visualization alongside results.
Users can build persistent tools like fitness trackers that pull weather data, calendar events, and location information into a single live dashboard. Google calls this "agentic coding in Search" powered by its Antigravity system.
Tom's Guide noted these mini apps launch initially in the U.S. for AI Pro and Ultra subscribers. The updated search box rolls out globally across desktop and mobile starting Tuesday. Users who want traditional link results can still select the "Web" tab.
Google also addressed the friction between AI Overviews and AI Mode. Asking a follow-up question in an Overview now redirects into AI Mode, the chatbot-like search experience.
Liz Reid, Google's vice president of Search, said during a briefing the company wanted to eliminate the "friction" between the two interfaces so users "don't have to think about where to go." On the Gemini side, Google introduced Spark, a 24/7 personal AI agent that works in the background even when a phone is locked or a laptop is closed. Spark handles recurring tasks like monitoring credit card statements and email inboxes, compiling summaries across Google Docs, Gmail, and Slides.
It goes into beta testing next week for AI Ultra subscribers, with Mac support and a phone monitoring feature called Android Halo arriving later. The Gemini app was rebuilt from the ground up with a new "Neural Expressive" design language that replaces the wall-of-text response format with bolded key information, fluid animations, and haptic feedback. A new Daily Brief feature rolls out today to U.S.
AI subscribers, pulling calendar, inbox, and task data into a prioritized morning digest that suggests next steps.
Google also launched Gemini Omni, an "any-to-any" video generation model available across all four AI plans, and introduced a cheaper $100-per-month AI Ultra tier aimed at developers alongside cutting the full Ultra plan from $250 to $200. The company said Gemini now has more than 900 million monthly users across 230 countries. Alphabet CEO Sundar Pichai said Google expects to spend about $180 to $190 billion this year on AI infrastructure and chips.
Anthropic accounted for 34.4% of paid AI business subscriptions in the U.S. in April, OpenAI held 32.3%, and Google was at 4.5%, according to finance platform Ramp. The gap explains the urgency behind Tuesday's announcements.
