Google has always treated the search box like a sacred object: small, familiar, and almost aggressively simple. For 25 years it trained people to think in fragments—type a few keywords, hit enter, scan blue links, repeat. But on Tuesday, Google effectively retired that old bargain. The company is redesigning the search box itself, not just what happens after you submit a query. And that distinction matters, because the interface you start with shapes the way you ask questions, the way you interpret answers, and the way you move through the web.
What’s rolling out now is more than a cosmetic refresh. Google is turning the search field into an AI-powered, multimodal conversation starter—one that can expand to fit longer, more detailed prompts and accept inputs beyond plain text. It also merges two previously separate AI experiences—AI Overviews and AI Mode—into a single continuous flow. In other words: Google wants you to stop thinking of search as a one-shot request and start treating it like an ongoing dialogue with the web.
This is the kind of change that looks incremental until you zoom out and realize it’s a shift in how the world’s most-used interface teaches users to communicate.
A search box that behaves like a conversation, not a command line
The most visible change is the search box’s new “shape.” Instead of nudging users toward short keyword strings, the field now dynamically expands to accommodate longer, more conversational queries. That sounds subtle, but it changes the psychology of searching. A narrow input area quietly encourages compression: you learn to boil your intent down to the smallest possible set of terms that might match what you’re looking for. A wider, more flexible input invites you to say more—context, constraints, follow-ups, and nuance.
Google is also adding multimodal input directly into the main entry point. Users can upload images, PDFs, files, and videos, or drag content from Chrome tabs right into the search experience. Previously, some of these capabilities existed in AI Mode, but they were gated behind additional steps. Now, the search box becomes the place where you bring your materials to the question, not just where you type a description of them.
That matters because multimodal search isn’t only about convenience. It’s about reducing the translation layer between what you have and what you want. If you’re trying to identify something in a photo, summarize a PDF, or reason about a video, the old model forces you to convert your input into text. The new model lets you keep the input closer to its original form, which generally improves both accuracy and user satisfaction—especially when the task requires interpretation rather than simple lookup.
Then there’s the “coaching” layer. Google is deploying an AI-powered query suggestion system that goes beyond autocomplete. Traditional autocomplete predicts the next word based on popularity and patterns. Google’s approach is meant to help users formulate complex, nuanced queries—essentially guiding them toward the kind of detailed questions that AI Mode handles best. This is important because many people don’t know how to ask good questions for AI systems. They may know what they want, but not how to express it in a way that yields useful results. A smarter suggestion system can reduce that friction and make the AI feel less like a tool you must learn and more like a system that meets you halfway.
The merged AI experience: fewer decisions, more continuity
If the redesigned search box is the front door, the bigger story is what Google is doing behind it. The company is unifying AI Overviews and AI Mode into one seamless search flow.
AI Overviews are the AI-generated summary panels that appear at the top of traditional search results. AI Mode is the more immersive conversational experience where you can ask follow-up questions and refine your understanding in a back-and-forth interaction. Historically, users had to choose which path to take—or at least navigate between experiences. Google’s new design removes that friction.
Starting Tuesday, a user can type a question, see an AI Overview alongside traditional results, and then continue directly into an AI Mode conversation to ask follow-up questions—all without switching into a separate interface. Google’s explanation is straightforward: the new AI search box is an upgrade of the traditional search box, so the results take you directly to main search rather than forcing you into AI Mode as a distinct destination.
For most users, that’s a big deal. Power users may actively seek out AI Mode, but the average person doesn’t want to think about whether they’re in “the AI experience” or “the normal experience.” They just want answers and the ability to refine them quickly. By merging the flows, Google is making the AI feel like a natural extension of search rather than a parallel product.
This is also a subtle bet about user behavior. When AI is presented as a separate mode, users may hesitate, delay, or abandon the experience if it feels unfamiliar. When AI is integrated into the interface people already trust, adoption becomes easier. Continuity reduces cognitive load, and reduced cognitive load is often the difference between a feature that’s impressive and one that becomes habitual.
Why Google is pushing this now: usage signals that the shift is already happening
Google didn’t announce this redesign in a vacuum. The company shared usage statistics suggesting that AI-driven search behavior is accelerating rapidly.
AI Mode, launched in the United States at I/O 2025, has surpassed one billion monthly users in its first year. Google says AI Mode queries have been doubling every quarter since launch. AI Overviews reach more than 2.5 billion monthly users. And overall search query volume hit an all-time high last quarter.
Those numbers are being used to support a specific argument: AI features are additive rather than cannibalizing. In other words, people aren’t abandoning search; they’re using it more, and they’re using it differently. Google’s leadership frames this as search becoming less about isolated queries and more like an ongoing conversation that connects users to deeper insights across the web.
There’s a deeper implication here too. If AI features are driving more search activity, then the interface becomes even more critical. When users search more often, they also interact with the entry point more often. The search box isn’t just a gateway—it’s the training ground for how people will learn to ask questions going forward.
Speed matters: Gemini 3.5 Flash as the engine for “feels instant” AI
A conversational search experience can’t be slow. Search is a high-frequency behavior, and users tolerate latency in some contexts but not in others. If the AI feels sluggish, the experience won’t scale—no matter how smart it is.
Google says the new search experience runs on Gemini 3.5 Flash. The company also upgraded AI Mode’s underlying model to 3.5 Flash, describing it as more powerful. Google claims Gemini 3.5 Flash outperforms its previous frontier model, Gemini 3.1 Pro, on nearly all benchmarks while running significantly faster in output tokens per second. The company positions it as near-frontier quality with dramatically lower latency.
This is the practical foundation for why the interface can work. A search box that invites longer, more conversational prompts increases the computational burden. Users are asking more complex questions, often with more context and sometimes with multimodal inputs. To keep the experience responsive, Google needs a model optimized for both quality and throughput.
In short: the redesign isn’t only about UI. It’s about making AI interaction feel as immediate as the old keyword experience—while being far more capable.
Search becomes a builder: generative UI and interactive visuals
One of the most interesting parts of Google’s announcement is that the search box is becoming a gateway to capabilities that go beyond returning text answers.
Google describes “generative UI,” which allows search to dynamically build custom widgets, interactive visualizations, and mini applications in real time tailored to the user’s question. The example Google gave is memorable: ask how black holes affect spacetime and receive an interactive visual in an AI Overview. Then follow-up questions can trigger new visuals generated on the fly.
This is a meaningful shift in what “search results” can be. For decades, search has been a list-based interface: titles, snippets, and links. Even when results include rich cards, the core structure remains navigational. Generative UI suggests a future where the result itself becomes an interactive artifact—something you can manipulate, explore, and use to understand the topic without necessarily clicking away immediately.
Google also previewed stateful, customizable experiences for ongoing tasks—planning a wedding, organizing a move, tracking a fitness routine. These are described as no-code, meaning users describe what they want in natural language and search builds the experience. Google says these will be available first for certain subscribers in the United States.
The unique take here is that Google is treating search less like a static retrieval system and more like a dynamic workspace. That changes the relationship between “finding information” and “doing something with it.” If search can maintain state and support ongoing tasks, it starts to resemble a lightweight application layer sitting on top of the web.
Information agents: search that watches the world for you
Another capability Google is introducing through this redesign is “information agents.” These are AI agents users can configure within search to monitor the web 24/7 for specific conditions and deliver synthesized updates when those conditions are met.
The example is straightforward: track market movements in a sector with defined parameters. The agent creates a monitoring plan, taps into real-time finance data, and proactively notifies the user when conditions are met—with links and context for further research.
Other use cases include apartment hunting, tracking sneaker drops, or monitoring any topic a user cares about. Google says information agents will launch first for certain subscribers.
This is where the redesign becomes more than a better search interface. It’s a step toward proactive assistance. Traditional search is reactive: you ask, it responds. Agents introduce a different rhythm: you set the conditions once, and the system keeps watching and returns when something changes.
That shift has major implications for how people use search. If agents handle monitoring and summarization, users may rely less on