Search Stopped Being a List And Started Being a Conversation
Google’s expanded search box signals a shift from keyword rankings to ongoing AI-driven conversations, where brands must monitor how they are represented across chat, agents, video, alerts, and off-screen AI interfaces.

Leela Adwani
AI Visibility Researcher and Editor
Update on
Visibility Monitoring

For twenty-five years, the deal with search was simple. You typed a few keywords, Google handed back a list, you picked a link. This week Google changed the deal. The search box now expands to take longer queries and chat-style exchanges, in what the company called the biggest change to the box since it launched. You no longer guess the keywords. You describe what you want, and you get an answer back.
That single change quietly rewrites two things every brand has been assuming about how it gets found.
The question got longer, and you can't see it coming
Keyword search trained everyone, brands included, to think in fragments. "Best running shoes." "CRM pricing." "Dentist near me." You could list those terms, track your rank on them, and roughly know where you stood.
A conversation doesn't work in fragments. When the box invites a full question, people ask full questions: messy, specific, loaded with context about budget, situation, and intent. There is no tidy keyword list behind "I run on pavement three times a week and my knees hurt, what should I buy." Google is leaning into exactly this, acknowledging that plain-language description beats keyword-guessing as a way to find things.
For a brand, the uncomfortable part is that you cannot pre-write the list of questions anymore. The surface where buyers describe their problem is wide open, and your brand either earns a place in the answer to a question you never anticipated, or it doesn't.
The answer stopped being a one-time event
Here's the part that's easy to miss: Google is putting agents inside search. "Agents in search is the next step," said Google DeepMind chief Demis Hassabis.
Concretely, that means standing queries. Instead of checking once whether your favorite band is touring, you set a query that alerts you whenever any of those acts books a nearby show. Google is extending the same idea, which it calls information agents, to recurring questions in shopping and news.
Sit with what that does to brand discovery. A one-time search was a single moment you could, in theory, win or lose. A standing query is a process that runs in the background and resurfaces answers to the same person, again and again, without them asking twice. If an agent is quietly monitoring a category on someone's behalf, it is repeatedly deciding which brands are worth surfacing. Get represented wrong once, and you get represented wrong on a loop.
The answer is also leaving the screen
The conversation isn't staying in the search box either. Google is pushing AI into nearly everything it makes:
Ask YouTube answers a how-to question with both text and a video that contains the answer.
The Gemini app is being built around agents for a wider range of tasks, with a new model and a personal assistant called Spark headed into Chrome and other apps.
Google is moving back into AI glasses, with an audio-only pair due this fall and a wearable called Project Aura in the works.
The thread running through all of it is that AI is becoming the layer between people and information, and that layer increasingly lives off the page, in a video, an app, or a pair of glasses. Hassabis framed the glasses as AI getting out into the real world. For a brand, it means the moment where you are described to a buyer can now happen somewhere you will never load in a browser.
What actually changes for brands
None of this is about a new tactic to chase; it's about a shift in where the truth about your brand gets decided.
When search was a list, you could audit your position on a known set of terms at a known moment. When search is a conversation, the inputs are unpredictable (you don't own the question) and the outputs are continuous (the answer keeps getting regenerated and re-served). A screenshot of what some assistant said about you last Tuesday tells you almost nothing about what it's telling a buyer with a slightly different question today.
The practical consequence is that point-in-time checking is over. The only way to know how your brand fares in a conversational, agent-driven surface is to watch it the way it actually behaves: continuously, across the assistants people use, against the real shape of the questions they ask.
The takeaway
The search box getting bigger looks cosmetic. It isn't. It marks the moment search stopped being a place you visit and became a conversation that runs with or without you, increasingly automated, increasingly off-screen. Brands spent two decades learning to rank in a list. The next job is harder and quieter: making sure that when AI speaks for you, in a chat, an alert, a video, or a voice in someone's ear, it gets you right.
Share on social media


