Google Just Replaced the Search Box With AI Agents. Here's What Changes for Enterprise.

For 25 years, the Google search box stayed the same: a white rectangle, a blinking cursor, ten blue links. At I/O 2026, Google replaced all of it. The box now expands dynamically, accepts images and videos alongside text, and feeds directly into AI Mode, which has passed 1 billion monthly users in its first year.

But the real announcement was not the box. It was what runs behind it: information agents that operate 24/7 in the background, scanning the web, reasoning across sources, and pushing updates to users without a single new query. Google did not redesign search. It turned search into an agent platform.

For enterprises already building and deploying AI agents, this changes the competitive surface in ways that matter right now.

What Google actually announced

Five changes shipped or were confirmed at I/O 2026. Each one pulls search further from the link-based model that enterprise content strategies depend on.

1. Gemini 3.5 Flash as the default

Google upgraded AI Mode with Gemini 3.5 Flash, which runs roughly four times faster than competing frontier models and outperforms the previous Gemini 3.1 Pro on coding and agentic benchmarks. Every AI Mode query now runs through a model built specifically for agent-style reasoning, not just retrieval.

2. The search box redesign

The biggest interface change in Google history. The new box expands to handle longer queries, suggests full questions (not just keywords), and accepts files, screenshots, video, and Chrome tabs as input. AI-powered suggestions anticipate intent before the user finishes typing. Queries in AI Mode already average three times longer than traditional searches, and this redesign is built to push that further.

3. Information agents

This is the headline feature. Users can now create background agents that monitor the web continuously for specific criteria. Looking for a two-bedroom apartment under $2,500 in Brooklyn with a dishwasher? Set up an agent. It scans listings 24/7 and notifies you when something matches. No repeated searches. No browser tabs left open.

Information agents launch this summer for Google AI Pro and Ultra subscribers in the US. But the architecture is the signal: Google is building persistent, goal-oriented agents inside its most-used product.

4. Personal Intelligence in 200 countries

Previously limited to US subscribers, Personal Intelligence now connects Gmail, Google Photos, YouTube history, and search history to AI responses for users in nearly 200 countries. No subscription required for most features. Search now knows what you have emailed about, what you have watched, and what you have searched for before. It uses all of that as context.

5. Generative UI and agentic booking

Search results no longer return static pages. For certain queries, Google generates custom interactive interfaces: dashboards, trackers, comparison tools. Combined with expanded agentic booking (now covering home repair, beauty, pet care, and local services), users can complete transactions directly in the search results page without visiting a business website.

The numbers behind the shift

The scale is already massive. AI Overviews serves 2.5 billion monthly users. AI Mode crossed 1 billion. Query volume in AI Mode has doubled every quarter since launch.

The impact on traditional search traffic is measurable. SISTRIX data from March 2026 shows click-through rates at position one dropping from 27% to 11%. Sixty percent of all Google queries now result in zero clicks. For news-related searches, that figure hits 69%.

Individual publishers are feeling it. HubSpot reported a 70-80% organic traffic decline. Chegg lost 49%. DMG Media saw drops of up to 89% on specific queries. NPR called it an "extinction-level event" for online news. Google's own search market share dipped from 92.9% to 89.6%, the steepest decline in the company's history, though that still leaves it dominant by any measure.

What this means for enterprise AI

Three implications stand out for companies building or deploying AI agents.

Google just validated the agent category at consumer scale

A billion people are now using AI Mode monthly. Information agents will put persistent, goal-oriented automation in the hands of everyday users this summer. The behavioral shift that enterprise AI platforms have been pitching for two years, where users delegate tasks to agents instead of doing them manually, is about to become a default expectation.

Enterprise buyers who were skeptical about whether agents are "ready" now have the largest tech company in the world shipping them to a billion people. That changes procurement conversations.

Your content strategy needs a rewrite

If your enterprise relies on organic search traffic for lead generation, the math has changed. Position one used to deliver 27% click-through. Now it delivers 11%. Zero-click searches are the majority. AI Overviews answer the query before anyone scrolls.

The companies that still get traffic are the ones getting cited inside AI Overviews. Being referenced by Google's AI is becoming more valuable than ranking below it. That requires a different content approach: authoritative, specific, data-rich content that AI systems want to cite as a source, not generic explainers that compete with 50 identical articles.

Agent-to-agent is coming faster than expected

Google's information agents run in the background, connect to external services, and complete transactions through the Agent Payments Protocol. This is the same multi-agent, tool-using, API-connected architecture that enterprise platforms like Beam have been building. The difference is that Google's agents will interact with your business whether you are ready or not.

Enterprises that have already deployed their own agents, built APIs that agents can call, and structured their data for machine consumption are positioned to work with this shift. Those that have not will find Google's agents bypassing them for competitors who have.

The bottom line

Google did not announce a search feature. It announced that the world's most-used information product is becoming an agent platform. 1 billion users in AI Mode. Background agents running 24/7. Personal context spanning email, photos, and browsing history. Transactions completing without a website visit.

For enterprises, the question is no longer whether to build AI agents. It is whether your systems, content, and data are ready for a world where your customers' first interaction with your business might be through someone else's agent.

The companies that treat this as an SEO problem will lose. The ones that treat it as an architecture problem, building agent-ready infrastructure, structured APIs, and AI-citable content, will win.

Ready to build AI agents that work in a world where Google Search is an agent platform?