{"id":15051,"date":"2026-05-25T15:14:39","date_gmt":"2026-05-25T12:14:39","guid":{"rendered":"https:\/\/www.stradiji.com\/?p=15051"},"modified":"2026-06-04T12:01:39","modified_gmt":"2026-06-04T09:01:39","slug":"google-i-o-2026-the-search-box-was-reborn-after-25-years-and-seos-new-name-is-selectability","status":"publish","type":"post","link":"https:\/\/www.stradiji.com\/en\/google-i-o-2026-the-search-box-was-reborn-after-25-years-and-seos-new-name-is-selectability\/","title":{"rendered":"Google I\/O 2026: The Search Box Was Reborn After 25 Years, and SEO’s New Name Is “Selectability"},"content":{"rendered":"

At the I\/O 2026 event held on May 19, 2026, Google announced the most fundamental transformation of the search box in the last 25 years. AI Mode reached 1 billion monthly active users and AI Overviews<\/a> 2.5 billion, while Gemini 3.5 Flash became the new default model. Search now accepts text, images, files, video, and even an open Chrome tab as input; information agents run 24\/7 in the background; Generative UI builds a custom interface for every query. Through a GEO lens, the picture is this: AI pulls 40\u201350 sources per query, narrows them to 12\u201320, and cites 3\u20138. For the SEO industry, one sentence sums it up: it’s no longer about ranking, it’s about being selected<\/b> as an input to the AI’s answer.<\/p>\n

The search box got its first major revision in 25 years<\/strong><\/h2>\n

In his opening keynote, Sundar Pichai announced that the Gemini app had reached 900 million monthly active users. But the number that really struck me came from the Search side: AI Overviews reached 2.5 billion monthly active users, and AI Mode hit 1 billion in just one year. Google described this as “the fastest product adoption we’ve ever seen.”<\/p>\n

The new AI-powered search box no longer accepts just keywords; it takes text, images, files, video, and open Chrome tabs together as input. In Search Engine Land’s words<\/i><\/a>, this is “the biggest change to the search box in 25 years.” The bar that has worked on a “type a word, get ten blue links” logic since 1998 has turned into a research assistant; on top of that, it now offers AI-powered question suggestions beyond autocomplete.<\/p>\n

Three main product lines: Gemini 3.5 Flash, Omni, Spark<\/strong><\/h3>\n

Gemini 3.5 Flash<\/b> is four times faster than rival frontier models; it became the default model of the Gemini app, AI Mode, and Spark. Gemini Omni<\/b> is a multimodal model that understands text, audio, image, and video simultaneously; it’s available free as a remix feature in YouTube Shorts. Gemini Spark<\/i><\/b><\/a> is Google’s 24\/7 personal AI agent, with Gmail, Docs, Canva, and Instacart integrations, offered in the new $100\/month AI Ultra plan.<\/p>\n

Information agents and Generative UI<\/strong><\/h3>\n

The quietest revolution of I\/O 2026 was information agents<\/i><\/a>. The user enters a request in natural language (“notify me if a 3+1 apartment comes up in \u00c7ekmek\u00f6y”), and the agent works in the background for days, scanning the web, news, price and stock data, and sends a notification when the criteria match. The paradigm shift for SEO: the user is no longer someone making a one-off query; your page is in a 24\/7 evaluation pool.<\/p>\n

Generative UI<\/b> designs the interface of the answer on the fly: an interactive simulation for astrophysics, a personal dashboard for finance, a 3D component diagram for how a watch works. Search builds this interface in the moment with Antigravity and Gemini 3.5 Flash. The result: the SERP is no longer uniform; the same query can return a different interface to a different user. The page you thought you “won” may appear on the user’s screen only as a component \u2014 or not at all.<\/p>\n

This also explains why the concept of “ranking” loses its meaning. LLMs are non-deterministic by nature; because of temperature settings, re-ranking layers, and real-time retrieval, they can give different answers to the same question at different times. Generative UI makes the interface variable on top of that. That’s why the new success metric in GEO isn’t “ranking” but citation probability<\/b>: how often you appear when you ask the same query 10 times across different sessions. There’s no fixed #1; there’s appearing in 7 out of 10 attempts.
\n<\/p>\n

\n
\n
\n <\/svg>\n <\/div>\n
\n

Found this helpful? Make us your go-to source on Google<\/p>\n

If our content has been useful, you can tell Google to show more of it when you search. Click below, then check the box next to Stradiji<\/strong> on the page that opens. It takes about five seconds.<\/p>\n

Add Stradiji on Google<\/a><\/p>\n

You’ll need to be signed in to your Google account. One click, then tick the box next to our name.<\/p>\n<\/p><\/div><\/div>\n<\/div>\n

AI Mode is now an ad surface<\/strong><\/h3>\n

One of the least-discussed but most directly SEO-relevant decisions at I\/O 2026: Google is placing ads inside AI Mode answers. This turns AI Mode from an experimental feature into a growing monetization surface for Google.<\/p>\n

The new ad formats are clear. Conversational Discovery Ads<\/b> answer a user’s question directly inside AI Mode; Gemini generates the ad creative for that specific query in real time rather than serving a static asset. Highlighted Answers<\/b> place sponsored results inside AI Mode’s list-style recommendations. On top of these come AI-powered Shopping ads<\/b> and Business Agent for Leads<\/b>, a smart brand agent you can chat with inside the ad itself. All formats carry a “Sponsored” label and include an independent Gemini-written explainer. Conversational Discovery Ads and Highlighted Answers are currently being tested in the U.S. on mobile and desktop.<\/p>\n

The SEO takeaway is clear. The AI Mode answer you’re trying to be cited in organically now also contains ads. So the visibility competition is no longer just with other brands \u2014 it’s with paid placement too. This doesn’t make organic GEO work pointless; on the contrary, as the paid space grows, being the organically “selected” source becomes more valuable. But set expectations correctly: AI Mode is no longer a purely organic surface.<\/p>\n

Ask YouTube and hardware: video SEO and local search are changing<\/strong><\/h3>\n

On the YouTube side, Ask YouTube<\/b> was announced. In response to complex queries, the AI finds the most relevant content across long videos and Shorts and produces a structured, interactive answer you can ask follow-up questions about. Two takeaways for video SEOs: video transcripts and timestamped chapter markers are now far more critical; and videos with crisp, bite-sized answer segments that “the AI can lift and use” will stand out.<\/p>\n

On the hardware front, new Android XR smart glasses will be on shelves this fall; partnerships with Warby Parker and Gentle Monster were announced. Live translation, Gemini-powered navigation, and notification summaries come as standard. For local SEO, glasses-based search is a whole new category: “what can I eat here” queries will now be asked out on the street, aimed at the storefront you’re looking at. Google Business Profile<\/a> maintenance, structured data<\/a>, and “near me” signals will take on a whole new importance over the next year.<\/p>\n

The GEO lens: what exactly does “selectability” mean?<\/strong><\/h3>\n

“Selectability” isn’t an abstract slogan; it’s a process with clear mechanics. AI search engines work with RAG (Retrieval Augmented Generation)<\/b>: when a user asks a question, the model splits it into 3\u20135 sub-queries (query fan-out), pulls 40\u201350 sources from the web, narrows them through mathematical fusion to 12\u201320, and finally cites 3\u20138 sources.<\/p>\n

The critical insight: your citation probability is directly proportional to how many times you appear in that initial 40\u201350 source pool.<\/b> Let’s lock it down with semantic triples:<\/p>\n