At Google I/O 2026, the tech giant unveiled a major overhaul of Search, shifting from a passive information gateway to an active, agent-driven platform capable of autonomous monitoring, booking, and custom outputs, signalling a new era in search-based interactions.
Google’s latest search overhaul marks a deliberate shift from retrieval to action. At its I/O 2026 developer conference on 19 May, the company said Search is moving into “the era of Search agents”, a change that folds autonomous behaviour into the product millions already use to find information, book services and write code. The significance is not just the features themselves, but the fact that Google is now distributing agentic tools through its dominant consumer interface, turning Search into a platform for work rather than a simple gateway to links.
The new approach rests on three elements. First are information agents, which Google describes as persistent monitors that can keep tracking topics across the web, social posts and live data feeds, then surface summaries and take certain actions on a user’s behalf. In practice, that could mean watching the housing market for a specific apartment profile or alerting a user when a suitable listing appears. Google said these agents will arrive first for AI Pro and Ultra subscribers in the summer rollout.
The second element is expanded agentic booking. Google Search now presents real-time pricing and availability for local services and experiences, while steering users to book with the provider they choose. In some categories, including home repairs, beauty and pet care, Google said it can even place calls to businesses for the user. That capability is being opened to all US users this summer, making Search a more active intermediary between discovery and transaction.
The third pillar is agentic coding inside Search. Google said Gemini 3.5 Flash and Google Antigravity now allow Search to generate custom visual tools, simulations, dashboards and trackers directly in response to a query. According to Google, the generative UI features will be available to all users in Search at no charge this summer, while more persistent custom experiences in Antigravity will follow later and begin with AI Pro and Ultra subscribers in the US. Google’s official Gemini 3.5 announcement says the Flash model is built for complex, agentic workflows, including sub-agent deployment and long-horizon tasks, and is available across the Gemini app, Search, Google AI Studio and Android Studio.
That model choice matters because it underpins the speed and scale of the system. Google says Gemini 3.5 Flash is the default for AI Mode worldwide, and chief executive Sundar Pichai claimed it is four times faster than rival frontier models while costing less than half as much. The company also said AI Mode has passed 1 billion monthly active users, AI Overviews has reached 2.5 billion users and Search queries have hit record levels. It put 2026 capital expenditure at roughly $180bn to $190bn, up sharply from $31bn in 2022, with the eighth generation of its custom TPU chips supporting that expansion.
For publishers and businesses, the wider consequence is structural. Search is no longer merely directing traffic; it is increasingly mediating the entire interaction through AI-composed surfaces, from summaries to booking flows to generated interfaces. That shifts visibility away from classic blue links and towards whatever the agent chooses to present. It also raises a basic question that Google’s own rollout does not answer: how reliable these actions will be once they are used at scale, and what standards of accountability will apply when an agent is monitoring, recommending or transacting on a user’s behalf.
Google’s expansion of Antigravity adds another layer to that shift. The company has framed the tool as part of a broader agent-development environment, with support for building and managing autonomous AI systems. Taken together, the announcements suggest that Google is trying to normalise agents not as a separate category of software, but as the default way users interact with Search. That may be the clearest sign yet that the market for autonomous systems is moving out of the laboratory and into the consumer mainstream, while the governance, trust and safety framework needed to support it is still catching up.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
