Google Search AI Mode adds agents, mini‑apps and Gemini 3.5 Flash

Google Search unveiled AI Mode upgrades that let the new Gemini 3.5 Flash model power “information agents” and on‑demand mini‑apps, turning the search box into a task‑oriented hub for all users worldwide【1】. The changes aim to shift Google Search from a query‑answer service to an execution layer that can monitor topics, build dashboards and even call local stores, a move that could reshape how businesses compete for visibility.

Expanded search box and AI‑driven inputs

The most visible change is a dynamically expanding search box that accepts not only text but also images, files, videos and even open Chrome tabs, delivering AI‑powered suggestions beyond traditional autocomplete【1】. Google says the box will surface AI Overviews—summaries that appear at the top of results—when users ask natural‑language questions, and a single click can jump straight into AI Mode【2】. This redesign is billed as the biggest update to Search in over 25 years, signaling Google’s intent to make the search interface the primary entry point for AI‑assisted tasks.

Information agents and on‑the‑fly mini‑apps

Google introduced “information agents” that run in the background, scanning the web, financial data, sports scores and social posts, then notifying users when relevant changes occur. The agents will first be available to Google AI Pro and Ultra subscribers this summer【1】【2】. In parallel, the Antigravity platform enables one‑prompt generation of custom UI, interactive visuals and mini‑apps—such as a wedding tracker or fitness dashboard—initially for paid subscribers in the U.S., with the generative UI feature opening to all users later this summer【1】. These capabilities echo Google’s “task‑based” vision, where Search not only returns information but also executes actions like price‑tracking hotels, calling local stores, or assembling travel itineraries【3】.

Competitive implications

By embedding agentic functions directly into Search, Google blurs the line between search and productivity tools, a space previously occupied by dedicated AI assistants and niche app builders. Competitors such as Microsoft’s Copilot and OpenAI’s ChatGPT plugins have offered similar task automation, but Google’s advantage lies in its native integration with Gmail, Photos and the broader Google ecosystem via Personal Intelligence, now expanded to nearly 200 countries and 98 languages without a subscription fee【1】. For marketers, the shift means that structured data and real‑time inventory signals will become critical for visibility, as AI agents will preferentially draw from well‑structured sources rather than traditional “ten blue links”【3】.

What to watch

• July 2026 rollout of AI agents for AI Pro/Ultra subscribers, the first phase of the task‑based vision.
• Release of generative UI to all users later in the summer, which could accelerate adoption of custom mini‑apps.
• Responses from rivals such as Microsoft and OpenAI, potentially accelerating their own integration of agentic features into search experiences.

Google’s AI Mode overhaul marks a decisive step toward a search experience that does more than retrieve information—it aims to act on behalf of users, turning the web into a real‑time workflow platform. The open question is whether the ecosystem of web publishers can adapt quickly enough to feed these agents with reliable, structured data.