From Blue Links to Generative UI
Google Search is shifting from a page of links into an interactive AI canvas. At its latest developer conference, Google introduced Google generative UI, a system that builds custom interfaces in real time in response to natural language queries. Instead of only returning text summaries or web pages, Search can now generate graphs, tables, simulations, and rich visual layouts tailored to the topic. Complex subjects like astrophysics or mechanical systems can be explained through interactive visuals rather than long-form text alone. The same generative UI tools can assemble dashboards for recurring tasks such as planning a wedding, moving home, or managing health routines. This moves Search beyond passive information retrieval and toward an adaptive environment where the interface itself is produced on demand, based on what a user is trying to understand or achieve.
Gemini 3.5 Flash and Antigravity Power New Gemini Search Features
Under the hood, the new interactive search results are driven by Gemini 3.5 Flash and a generative UI engine called Antigravity. Gemini 3.5 Flash is now the default model in AI Mode globally, enabling faster, more conversational responses across text, images, files, videos, and even Chrome tabs as inputs. The redesigned AI-powered search box dynamically expands as questions become more detailed, offering AI suggestions that go beyond traditional autocomplete. Antigravity, meanwhile, acts as a generative layout engine that can assemble mini-app-like experiences using live data such as maps, reviews, and weather. Users can end up with bespoke tools and trackers that feel closer to lightweight applications than standard web results. Together, these Gemini search features blur the line between a search query and an on-the-fly app build, with the model deciding not just what to show, but how it should look and behave.
AI Search Agents Turn Queries Into Ongoing Tasks
Beyond single-shot answers, Google is introducing AI search agents that continuously work on a user’s behalf. These agents can monitor blogs, news sites, social platforms, and real-time feeds covering areas like finance, shopping, and sports. Users define requirements—for example, a budget range, preferred neighborhoods, or product specifications—and the information agent tracks relevant updates over time. In practice, that means ongoing tasks such as apartment hunting or product scouting can be managed directly from Search, with alerts and curated options surfaced automatically. Google is also extending agentic booking tools to local services and experiences. People will be able to specify dates, constraints, and preferences, then see pricing and availability and follow links to complete the booking. In some categories, users can even ask Google to call businesses on their behalf, turning Search into an active coordinator rather than a static results page.
A New Mode: Search as Simulation and Workspace
The combination of generative UI and AI search agents pushes Search into a new role: a simulation and task workspace that lives inside the browser. Instead of bouncing between sites and tools, users can remain in AI Mode while refining plans, validating options, and tracking progress. Dashboards generated by Antigravity can act as living documents that update with new information from the web. Shopping flows are being tightened through Universal Cart, which lets people add items while using Search or Gemini and then monitor price changes, stock alerts, and compatibility. Upcoming protocols for agent-based payments and commerce aim to let AI agents check out under user-defined limits while keeping merchants as the seller of record. All of this positions Google Search as a programmable environment where queries trigger not just information retrieval, but evolving, interactive workflows.