Google Search's biggest change in 25 years: an intelligent box that understands, acts, and plans

Google Search is changing from a place where people type keywords into a place where people describe tasks. At I/O 2026, Google called the new AI-powered Search box the biggest upgrade to the Search box in more than 25 years. That is not only a design claim. It signals a product shift from search engine to task engine.

The new Search box can dynamically expand, suggest richer actions, accept more types of input, and move users into conversational AI Mode. It is also connected to a broader set of features: information agents that monitor topics in the background, booking and calling tools for local services, generated interfaces with tables and simulations, and persistent dashboards or mini apps for ongoing tasks.

This article is not about how websites should adapt to AI answers. It is about how the Search box itself is becoming a workspace for planning, monitoring, comparing, and acting.

For the model behind this shift, see Truescho's separate article on Gemini 3.5 at Google launches Gemini 3.5: did the era of assistants that act, not just answer, begin?. For Google's return to wearable assistants, the smart glasses analysis is at Google smart glasses are back: will they succeed after Google Glass failed?.

What changed in the Google Search box

Google says the new intelligent Search box starts rolling out on May 19, 2026 in all countries and languages where AI Mode is available. AI Mode uses Gemini 3.5 Flash as its new default model globally, connecting the Search interface to Google's latest agentic model family.

The box is no longer just a short input field. It can expand dynamically and provide AI-powered suggestions beyond classic autocomplete. Users can bring in text, images, files, videos, and Chrome tabs. Follow-up questions can move from a generated answer area into conversational AI Mode across desktop and mobile worldwide.

That is a major behavioral change. Classic Search starts with a query. The new Search box can start with a goal:

1. "Compare these three universities for a master's in data science."
2. "Use this PDF and this website to build a checklist."
3. "Monitor price drops for this laptop and summarize changes."
4. "Find a pet groomer near me and call to check availability."
5. "Create a dashboard for my scholarship deadlines."

The Search box becomes a control panel. The search result becomes one possible output, not the whole product.

The rollout map matters

Google's announcement includes several features, but they do not all arrive for the same users at the same time. Readers should separate global Search box changes from subscription-only agents and U.S.-limited booking features.

This table is the most important part of the announcement for real users. A student in the Philippines may receive the new AI Mode experience where supported, but not the U.S. booking calls. A researcher in Germany may get generated interfaces later in Search, while mini-app dashboards first start with U.S. Pro and Ultra subscribers. A shopper in Texas may see booking and calling sooner than someone in Brazil.

Multimodal search becomes normal

Search used to begin with words. The new box accepts a broader set of signals: text, images, files, videos, and Chrome tabs. That changes what a "query" means.

A user can ask about a chart from a PDF, a product photo, a lecture recording, or a set of open tabs. Instead of forcing everything into keywords, the user can provide the actual object they are working with. This is especially useful for students, researchers, travelers, and shoppers.

Consider a medical student in Manila reviewing three lecture PDFs, two YouTube explainers, and a scanned diagram. The old search pattern would require many separate searches. The new pattern could ask Search to compare the materials, identify contradictions, and produce a study plan. The result is less like a list of links and more like a working session.

Truescho readers can apply the same idea to education planning. Start with Truescho opportunities, compare institutions through Truescho rankings, calculate target grades with the GPA calculator, then use the Search box to organize the next steps.

Information agents: Search that keeps watching

One of the biggest changes is the arrival of information agents. Google says these agents can work in the background 24/7, monitor the web, blogs, news, social sources, and fresh finance, shopping, and sports data, then send synthesized updates with possible actions.

This is different from setting a simple alert. A traditional alert tells you that something new appeared. An information agent is supposed to interpret changes, connect them to your goal, and propose next steps.

For example, a Brazilian student named Rafael wants a funded master's program in renewable energy. He tracks 18 universities in Europe and Canada, has a monthly budget limit of EUR 1,100, and needs an application deadline after November 15. Instead of searching every Friday, he could ask an information agent to watch scholarship pages, university news, and tuition updates. If three programs change deadlines and one adds a new funding page, the agent could summarize the changes and suggest which applications deserve attention.

If Rafael saves 2 hours per week across a 16-week application season, that is 32 hours recovered. The value is not only time saved. It is reduced missed opportunity risk.

The limitation is access. Google says information agents launch first for Google AI Pro and Ultra subscribers in summer 2026. That means they are not a universal free Search feature at launch.

Booking and calling: Search takes action in the local world

Google also says booking expands to local experiences and services. For selected categories such as home repair, beauty, and pet care, Google can call businesses on the user's behalf. These booking and calling capabilities are set to roll out to everyone in the U.S. in summer 2026.

This is a major change because it moves Search from answering "who is available?" to helping complete the task. A user might ask for a dog groomer within 5 miles, under $80, with availability after 5 p.m. The assistant could find options and call to confirm details, depending on the category and rollout.

The international limit is important. Users outside the U.S. should not assume calling features are available to them in summer 2026. Google may expand later, but the official rollout described in the research is U.S.-first for these booking and calling capabilities.

Generated interfaces: Search results become tools

Google says Search can generate custom UI: layouts, interactive visuals, tables, graphs, and simulations. These generated interfaces are expected for everyone in Search in summer 2026, free of charge.

This could be one of the most useful changes for complex questions. Some answers are best shown as a paragraph. Others need a comparison table, a timeline, a calculator, a map, a chart, or a slider. If Search can generate the interface that fits the task, the result page becomes more like a temporary app.

For students, this could mean a generated scholarship comparison table with tuition, living costs, deadlines, language requirements, and documents. For a traveler, it could mean an interactive route plan. For a small business owner, it could mean a cost simulator. For a sports fan, it could mean a live standings dashboard.

The power is in matching form to purpose. A paragraph is not always the right answer.

Mini apps and dashboards: the Search box as a personal control room

Beyond one-time generated interfaces, Google says Search can build persistent custom dashboards, trackers, and mini apps for ongoing tasks. These start first for Google AI Pro and Ultra subscribers in the U.S. in the coming months.

This points to a new kind of Search memory. A one-time query answers the moment. A dashboard follows the project.

A dashboard does not need to replace dedicated apps. Its value is speed. If a student or researcher can create a temporary workspace in Search for a 3-month project, they may not need to build a spreadsheet from scratch.

Personal Intelligence and private context

Google says Personal Intelligence expands in AI Mode to nearly 200 countries and territories across 98 languages, with no subscription required. It includes user-controlled connections to Gmail, Google Photos, and soon Calendar.

This is powerful because personal context makes answers more useful. Search can understand your trip, your documents, your photos, or your schedule. It is also sensitive because those sources contain private information.

The safer approach is selective connection. Do not connect everything by default. Use personal context when the benefit is clear, such as summarizing a travel itinerary from Gmail or finding a receipt. Review permission settings. Disconnect sources when a project is finished. Treat AI-assisted Search like a helpful assistant sitting next to your inbox, not like an invisible background process that should see everything forever.

A realistic example: Amara in Toronto

Amara is a Nigerian graduate student in Toronto applying for a PhD and managing a part-time job. She tracks 22 potential supervisors, 9 funding pages, 6 conferences, and 4 visa-related deadlines. Her spreadsheet has 147 rows after two months.

With the Google AI Search box 2026 experience, her workflow could change. She could upload a funding PDF, paste a supervisor page, ask Search to compare fit with her research statement, then move follow-up questions into AI Mode. Later, an information agent could monitor selected university pages if she has access to the required subscription tier. A generated interface could turn the results into a deadline tracker.

Amara still verifies every official page before submitting. She still writes the final emails. But if the Search box reduces her weekly research maintenance from 4 hours to 90 minutes, she gains about 10 hours per month. For students balancing work and applications, that is significant.

This is also where a platform like Truescho fits naturally. Truescho can help users find global opportunities and compare study options, while Google's Search box can help turn those sources into personal plans. The two roles are different: one gathers education resources; the other helps operate on the user's task.

What not to assume

The announcement is large, and it is easy to overstate it. Several boundaries are clear:

1. The intelligent Search box starts where AI Mode is available, not necessarily every market in the same way.
2. Information agents first launch for Google AI Pro and Ultra subscribers in summer 2026.
3. Booking and calling capabilities roll out to everyone in the U.S. in summer 2026, not globally.
4. Mini apps and dashboards start first for U.S. Pro and Ultra subscribers.
5. Personal connections to Gmail, Photos, and Calendar are user-controlled and should be treated carefully.

There is also a trust question. If Search acts on your behalf, the cost of misunderstanding rises. A wrong search result wastes time. A wrong booking or incorrect call can waste money, damage relationships, or create scheduling problems. Google is moving Search toward action, but users need confirmation points.

How students and researchers can use it well

The best use cases are structured, reviewable, and source-based. Instead of asking broad questions like "find me the best scholarship," users should provide constraints and ask for a table:

1. Country preferences
2. Degree level
3. Budget ceiling
4. Language test status
5. Deadline range
6. Required documents
7. Source pages to compare

Then they should ask the Search box to identify missing information and link back to sources. For grade planning, use a reliable calculator such as the Truescho GPA calculator. For discovery, start with Truescho opportunities. For institution comparison, use Truescho rankings. AI becomes most valuable when it operates on a clear workflow, not when it guesses from a vague prompt.

The bigger meaning: Search becomes a workspace

Google's biggest Search change is not that answers become longer. It is that Search starts to behave like a workspace. It can understand richer inputs, carry a conversation, monitor changes, call businesses, generate interfaces, and build persistent tools.

That will not eliminate the need for websites, apps, spreadsheets, calendars, or expert advice. It will change the starting point. Many tasks that used to begin with a search result page may now begin with an AI-powered workspace inside Search.

For everyday users, the practical rule is simple: use the new Search box for comparison, planning, monitoring, and first drafts. Use official sources for final decisions. Use human judgment for anything that affects money, health, immigration, employment, grades, or legal status.

FAQ

What is Google's new AI-powered Search box?

It is a redesigned Search input that can expand dynamically, suggest richer actions, accept text, images, files, videos, and Chrome tabs, and move users into conversational AI Mode. Google calls it the biggest upgrade to the Search box in more than 25 years.

When will the new Google Search box roll out?

Google says the intelligent Search box starts rolling out on May 19, 2026 in all countries and languages where AI Mode is available. Other features have separate timelines, including information agents and booking tools in summer 2026.

What inputs can the new Google Search box understand?

Google says users can bring text, images, files, videos, and Chrome tabs into the new Search experience. This allows Search to work with real materials such as PDFs, product photos, open research tabs, and video content instead of only typed keywords.

What are information agents in Google Search?

Information agents are background agents that can monitor the web, blogs, news, social sources, and fresh finance, shopping, and sports data. They synthesize updates and may suggest actions. Google says they launch first for Google AI Pro and Ultra subscribers in summer 2026.

Can Google Search book appointments or call businesses?

Google says Search booking expands to local experiences and services, and for selected categories such as home repair, beauty, and pet care, it can call businesses on the user's behalf. These booking and calling capabilities roll out to everyone in the U.S. in summer 2026.

What are generated interfaces in Search?

Generated interfaces are custom layouts, interactive visuals, tables, graphs, and simulations created by Search for a specific task. Google says this capability is expected for everyone in Search in summer 2026, free of charge. It turns some results into temporary tools.

Can Google Search build mini apps or dashboards?

Yes, Google says Search can build persistent custom dashboards, trackers, and mini apps for ongoing tasks. These start first for Google AI Pro and Ultra subscribers in the U.S. in the coming months, so they should not be assumed globally available at launch.

Is the new Search available outside the U.S.?

Parts of it are. The intelligent Search box starts where AI Mode is available, and Personal Intelligence expands to nearly 200 countries and territories across 98 languages. However, booking and calling are U.S.-first, and mini apps start with U.S. Pro and Ultra subscribers.

Sources

• Google: Search at I/O 2026
• Google: Sundar Pichai I/O 2026 keynote post
• Google: Gemini 3.5 official announcement
• The Verge: Google coverage
• TechCrunch: Google I/O coverage