At Google I/O 2026 the company announced one of the most significant changes to Search in decades: the search box is being redesigned, AI Mode gains new capabilities, information agents are introduced, and generative in‑search interfaces and mini‑apps will appear. The details below are summarized by Szuhi Attila, founder and editor‑in‑chief of ITE.hu.
What is changing in the search box?
According to Google, the search box receives its largest update in more than 25 years. The new AI‑enhanced search box no longer serves only for typing keywords: it dynamically expands, helps users refine queries, and accepts multimodal inputs — text, images, files, videos and even open Chrome tabs. Google said the new interface will roll out in countries and languages where AI Mode is available; it is already active in Hungary.
Behind the scenes, the default model for AI Mode will be Gemini 3.5 Flash worldwide, a model also introduced at the conference. Google positions this model as particularly strong for agentic tasks and coding, which matters because Search is shifting from giving answers toward executing tasks, monitoring data, assisting bookings, creating visual content and even generating mini applications.
From keywords to context: input is changing
The major shift is not simply adding an AI box to Search, but changing what users can provide as input. Previously SEO largely assumed a user types a query like “best coffee maker” or “rain jacket for hiking.” Under the new model users can attach images, documents or open tabs, making intent far more complex.
For example, a user could upload a photo of their living room, supply dimensions and preferred stores, then ask: “find similar light‑colored sofas suitable for a small apartment under 250,000 forints.” That is no longer a conventional keyword search but a full task.
From an SEO perspective content must be interpretable in context: product data, images, videos, specifications, comparisons, reviews and situational usage details grow in importance.
AI Overview and AI Mode become a process
Google also said it will make it easier to move from AI Overview to AI Mode: after an Overview the user can ask follow‑up questions in AI Mode while the context remains. This turns searching into a multi‑step process — general question, AI Overview, deeper follow‑up in AI Mode, filtering, comparing and then decision or purchase.
SEO therefore should consider not only the first query but how a page can appear at later stages of a longer decision process.
Information agents: Google watches and notifies
A major introduction is information agents (AI agents). According to Google, these agents can continuously monitor the web — news sites, blogs, social posts, price or sports feeds — and notify users when relevant changes occur. Initially these features will be available this summer to Google AI Pro and Ultra subscribers.
These agents go beyond simple keyword alerts: they interpret, filter, summarize and may suggest actions. Sites that publish new, clear and trustworthy information quickly could gain advantages — including news outlets, price and product trackers, real estate portals, job boards, financial and sports data providers, local service listings and e‑commerce sites.
Booking, calling, buying: search moves closer to conversion
Google plans to extend agent capabilities from information to actions: agents may be able to book local experiences and services, compare prices and availability, and in certain categories — such as home repair, beauty services or pet care — even place calls to businesses on the user’s behalf. These capabilities will appear first in the United States.
This can materially affect local SEO. Historically the goal was often to drive the user to a website to make contact or fill a form. Under the new model Google can handle more steps inside Search. That does not render websites useless, but their role may change and agents, not only users, will access them. Accurate, consistent Google Business Profile data, structured data, opening hours, pricing, service descriptions and reviews become critical. Google’s generative AI SEO guidance also highlights Merchant Center and Google Business Profile data as ways to surface products and local services in AI answers and other Search surfaces.
Generative interfaces and mini apps in Search
Another important announcement is generative interfaces: Search can generate visual tools — charts, tables, simulations or interactive elements — tailored to a query. Google says these generated surfaces will be available to all users for free this summer.
Further, the Antigravity integration could enable Search to create custom mini‑apps — for example a fitness tracker or wedding planning dashboard — rolling out first to Google AI Pro and Ultra subscribers in the U.S.
For SEO this means Google’s own surfaces could replace certain calculators, comparison tools or simple interactive solutions that previously drove traffic. At risk are low‑value calculator pages, generic definition content, basic comparison lists and templated “best of” articles; stronger will be pages offering original data, deep expert analysis, community feedback, unique tests or proprietary viewpoints.
Traffic may fall, but not all clicks are equal
Within the SEO community there is concern that AI answers, agents and mini‑apps will reduce clicks to websites. Google stresses that AI features will still show links and may surface more diverse sources via query fan‑out. In practice the effect will likely be mixed: simple informational queries may produce fewer clicks because answers appear on Google’s surface, while more complex, decision‑oriented queries may yield higher‑quality clicks because users arrive better informed.
Google’s documentation claims AI Overviews can lead to more engaged visits — for instance more pages viewed per session.
Multimodal search reshapes product SEO
A key e‑commerce implication is that when users can submit an image or screenshot, product discovery no longer relies solely on text keywords. Visual similarity, appearance in context and the accuracy of product data gain importance.
Webshops should therefore move beyond keyword‑only descriptions: high‑quality product imagery, multiple views, real life usage photos, precise product attributes (size, material, color, compatibility), structured product data, inventory and pricing info, reviews and comparable specifications will matter more.
Search AI will need to understand what a product looks like, what it is for, who it suits, how it differs from alternatives and in what contexts it should be recommended — not merely whether the page contains a specific phrase.
Likely biggest SEO impact
The most significant change is that Search will become less linear: users will not simply type a keyword, scan a list of results and click. Instead they will converse, upload, refine, compare and often proceed toward a decision within Google’s surfaces.
Three implications:
- Simple informational traffic may decline further.
- Complex, decision‑near content may become more valuable.
- Visibility will be broader than classic organic rank.
SEO will not disappear, but ranking well on a single result list will be less sufficient. The critical question becomes whether Google’s AI systems recognize a site as a useful, reliable, contextually relevant source. The winners will be sites that offer what the AI cannot easily replicate: products, services and unique primary data.
Practical takeaway
Publishers and merchants must focus on quality, context and accurate structured data. The challenge is shared by competitors: those who adapt faster can gain advantage. Notably, Google did not make AI Mode the universal default, which allows gradual adoption.
How quickly users will adopt the new tools and how precisely behavior will change remains uncertain.
Author: Szuhi Attila ITE.hu
Referenced technologies and services
- AI Mode (default model: Gemini 3.5 Flash)
- AI Overview
- information agents (AI agents)
- Antigravity integration (mini apps)
- Google AI Pro and Ultra subscriptions
(Summary based on Google I/O 2026 announcements.)
