Google's search box is now a task dispatcher

Yesterday Google said it was the biggest change to the search box in over 25 years. That part is true, and not for the reason most of the coverage is fixating on. The interesting thing is not the new model or the conversational interface or the agentic booking. It is what the search box has quietly become underneath all of that.

It is no longer a query receiver. It's a task dispatcher.

TL;DR: The unit of search shifted from query to task this week. The SEO inputs you control (crawlability, structured data, named expertise, GBP, entity clarity) matter more now, not less. The "SEO is dead" headlines are misreading the AEO guide Google shipped alongside the announcement. Audit your AI-citation eligibility, verify your schema actually renders, and stop paying for the deliverables Google just disowned.

I read the Google blog post and the AEO guide back to back for two days. The conclusion I keep arriving at is the opposite of the headlines. I might be wrong about the long arc (nobody has 2027 data) but the near-term reading feels solid.

Side-by-side diagram comparing the old SEO funnel of query, ten blue links, click, read with the new task-dispatcher funnel of conversational input, AI Mode synthesis, generative UI layout, agent follow-up over time. — Old funnel was one query, one decision. New funnel is one task, many touches, sometimes ambient.

From query to task

Last month I started looking for an apartment in Berlin. Two-bedroom, under €1,500, close to a Ringbahn station so my partner can reach her studio without changing trains. Under the old model the workflow was familiar. I type "apartment Berlin Ringbahn 1500 zwei zimmer" into Google, get ten blue links, click four listing aggregators, mentally merge the inventory, repeat with different filters, and by Friday I have thirty tabs and a spreadsheet.

I tried the same task in AI Mode this week, in plain English. The model surfaced seven candidates with commute time to my office and rent comparables per Bezirk, then offered to keep watching the saved task. The Information Agents Google announced for AI Pro and Ultra this summer will do exactly that. Google describes them as operating "in the background, 24/7," reasoning across the web to monitor for changes. (I'm quoting loosely; the exact phrasing is about agents that "intelligently look across everything on the web," but the 24/7 background language is Google's own.)

Four announcements ladder up to the same shift. Gemini 3.5 Flash is now the default model in AI Mode globally. Information Agents arrive for AI Pro and Ultra this summer. Agentic booking, where Google completes a reservation on your behalf, rolls out to everyone in the U.S. this summer for local services and experiences. And generative UI, the system that composes a custom layout for your query in real time, ships free for all users this summer too.

Put those together and the apartment search stops being one query I fire on Tuesday. It becomes a session that runs for two weeks while I sleep, work, and forget about it. That is the whole thesis, so I'll state it once: the unit of search is now the task, and the box that opens it is a dispatcher. Everything below is what that does to the work.

From pull to push

The user used to fetch. They typed, they clicked, they read. The agent now fetches on their behalf, sometimes while the user is asleep. That sounds like a small UX wrinkle. It is the part that quietly rewrites what "search volume" even means.

Picture one Information Agent watching one saved task for thirty days in your GSC. Zero impressions for the user. Zero clicks. The agent did the work; the user got a single notification at the end. Multiply that across a billion monthly AI Mode users (Google's number) and the demand curve changes shape. I want to be careful here: this is a plausible mechanism, not a measured one. Google hasn't published agent-level query data, so "ambient demand GSC can't see" is an inference, not a confirmed effect.

What we can measure is the gap between two curves. Google said AI Mode passed one billion monthly users a year after launch, with queries more than doubling every quarter since. On our own dashboard across the last 90 days, AI Mode citation count grew faster than impressions did, and that gap is the whole story. (Caveat: one company, one quarter, and I can't fully separate the citation growth from our regular SEO work.) If you're still measuring SEO by impressions alone, you're staring at the smaller number. We wrote about the click collapse last year; it hasn't reversed, it has compounded.

From static page to generative UI

One sentence in the Google I/O post deserves more attention than it got: Search now builds custom layouts in real time, assembling them from components on the fly. For free users. This summer. (I'll hedge the mechanism, not the fact: how aggressively it replaces the classic results page, and for which queries, nobody outside Google can size yet.)

That reframes the page as a raw material instead of a destination. The user is not seeing your page. They're seeing a composed component that pulled three sentences from your page, a price from a competitor, a review snippet from Reddit, and a map block from a Google Business Profile. Your page contributed an ingredient. It did not serve the meal.

Two-column before-and-after table showing the seven paradigm shifts: query to task, pull to push, static page to generative UI, page to entity, keyword density to entity clarity, link velocity to named expertise, LLMs.txt to indexability. — The seven shifts that matter. Left column is the input most teams optimized for through 2024. Right column is what those inputs roll up to now.

Click-through rate gets hard to read in this world. A page can be cited five times in a generative surface and earn zero clicks; that same page two months prior would have ranked #3 and earned the click. The page got better at its job and the metric got worse. That's the trap: people watch the dashboard crater and conclude the work failed, when the work is the only thing that put them in the layout at all.

Here's the whole shift in one view.

Dimension	Old model (pre-AI Mode)	New model (task dispatcher)	Operator lever
Unit of demand	Single query → single SERP	Task → multi-touch session spanning hours or days	Entity clarity so the same source gets cited across session follow-ups
How the user receives content	Clicks a blue link, lands on your page	Reads a generative layout assembled from several sources; may not click at all	Structured data (FAQPage, Product, LocalBusiness) so components pull from your page, not a competitor's
When search runs	User-initiated, synchronous	Agent-initiated, asynchronous, sometimes while the user is offline (Information Agents)	Crawlability and indexability; if the agent can't read your page at 3 AM, it picks another source
Primary success metric	Click-through rate, rank position	Citation count, AI-mode citation share	Named-author expertise and verifiable bios; anonymous content is a citation handicap
What "SEO work" means	Keyword targeting, link velocity, on-page variants	Entity consolidation, clean schema, page experience, GBP completeness	The AEO guide's "do" list: indexability, semantic HTML, Business Profile, Merchant Center

Read down the fourth column. Every lever there is something you already controlled before this week.

From pages to entities

The AEO guide hints at something Google has been building toward for years. It has indexed URLs that contain words for 25 years; increasingly it reasons about businesses, products, and services as entities across many sources. A page is one signal among many for the entity it represents, which is why Google calls out Business Profiles and Merchant Center by name in its "still matters" list. Those are entity surfaces, not page surfaces: they tell Google what your business is, not what one of your URLs says. The "you don't need" list is the mirror image, all page-level deliverables that fragment your entity signal: LLMs.txt files, content chunking, special schema beyond what's supported. Consolidate the entity, don't split it.

In practice that means one string everywhere. Same business name across homepage, GBP, citations, and schema. Same address format on every surface. Same services and categories (if your GBP says "marketing agency" but your homepage h1 says "demand generation consultancy," you're blurring your own entity). Named experts with real bios and sameAs links. Consistent product specs across GBP, Merchant Center, and your PDPs. Our GBP guide and our piece on local search intent both got more useful this week, not less.

Why SEO matters more, not less

This is the part I want to nail, because the doomer takes are loud and wrong on the evidence. Google shipped a developer document this week titled "Apply foundational SEO best practices to generative AI search." Read the heading twice, then the load-bearing line: "The best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems." Google is telling you, in its own docs, that the foundations didn't move. The headline writers skipped that paragraph.

Which inputs carry more weight now:

Structured data that parses cleanly: FAQPage, Product, LocalBusiness, Person. The model uses it to ground entity reasoning
Named-author expertise with verifiable bios and sameAs links; my read is that the model surfaces named sources more often than anonymous ones, though Google hasn't quantified the gap
Crawlable, indexable, JS-rendered content. The AEO guide notes browser agents may inspect the DOM and the accessibility tree, which makes clean semantic markup an agent input, not just a screen-reader input
Entity clarity and good page experience, because an agent reads your page closer to the renderer than to a human visitor

Which inputs carry less weight: keyword density and on-page variants (the model resolves synonyms), AI-spun thin content (the model has its own), link velocity tactics, and the AI-SEO deliverables Google just disowned by category. If you're cutting a check for any of them, you're paying for work Google says adds no signal. (Other AI providers may behave differently; I'm talking about Google's surfaces here.)

A dogfood note before I move on, because I'd rather you learn this from my mistake than your traffic. This past week we pushed FAQ JSON-LD to three articles on this blog. The CMS escaped the <script> tag on save, so the live HTML shipped the schema as plain text and Google read zero FAQ markup. We only caught it by view-sourcing the page. Structured data that does not parse is worse than none: it costs you opportunity without telling you why. Our piece on AI Overview citations walks the diagnostic step by step.

Where the "Google Zero" headlines are right

Doom-mongering is off-brand for us, but pretending nothing changed is dishonest, and dishonest pieces age badly. The HouseFresh case is real. Gisele Navarro discussed it on Decoder in May 2024, and HouseFresh's own write-up documents a traffic collapse in the 90 to 95% range on a high-quality independent air-purifier review site. (I had the number wrong in an earlier draft; the ~65% figure I'd seen floating around was a different publisher's ad-revenue drop, not HouseFresh's traffic.) It's happening to other review sites too. Nilay Patel's "Google Zero" framing was directionally correct for a specific cluster of content types.

Content shapes the AI Overview and AI Mode synthesis drain:

Product comparison listicles: "best mattresses 2026", "top 10 CRM tools"
How-to content that fits inside a 300-token AI Overview summary
Definition-style explainers; the model has its own answer for "what is a webhook"
Roundup posts that aggregate existing sources without adding primary research
Thin affiliate review content that reuses manufacturer copy

Content shapes still defensible, and in some cases gaining value:

Original reporting with named sources
Primary research with proprietary datasets the model can't scrape
Named expertise on narrow topics
Transactional and local pages tied to a real physical business
Non-copyable artifacts: live tools, calculators, datasets, interactive visualizations

So "is SEO dead" is the wrong question; the answer depends entirely on what you make. For listicle content, the easy money is gone. For original work tied to real expertise or a real business, the value went up.

One corollary worth naming, because it's the part that scares me most. The same shift that makes good content matter more also makes bad content matter more, since the AI picks a single answer from a small set of sources. A planted page can flip what AI says about your brand fast, and you may not notice until a customer quotes it back to you. The defender's playbook is in the companion piece on AI-answer ownership.

What an operator can control this week

Concrete and fast. Five moves, in order.

Audit your AI-citation eligibility. Sample thirty of your most important URLs. Run each target keyword through AI Mode. Note which got cited and which got skipped. That's a baseline you can compare against in 30 days. Our checker at /tools/ai-visibility-checker/ runs it without a login.

Venn diagram with two circles. Left circle labelled 'You used to control': keyword targeting, page-level CTR, SERP position, link velocity. Right circle labelled 'You still control': crawlability, structured data, named expertise, entity clarity, GBP completeness, JS-rendered content, page experience. Intersection labelled 'Quality + intent' contains both old and new inputs. — What an SEO operator still controls in 2026. The shaded overlap is where the AEO guide says the same fundamentals still apply.

Verify your structured data renders as executable script. View the live source on any page that should ship FAQ or Product schema. Search for <script type="application/ld+json". If it's there, your CMS is silently stripping the schema, exactly the bug we hit last week. Our audit tool flags it for you.

Name your authors and ship author schema that parses. Real bios, real photos, real sameAs links to LinkedIn and X. The AEO guide is explicit that expertise signals still feed the systems generative AI sits on top of. Anonymous content is a citation handicap, not a neutral choice.

Tighten entity clarity. One audit pass: business name, address, services, categories, the same string across GBP, homepage, About page, schema, and citations. The GBP guide walks the full checklist.

Stop spending on the cottage industry Google just disowned. LLMs.txt files, content chunking, AI-specific rewriting, inauthentic mention packages. Any open invoice for those is the easiest budget reallocation you'll make this quarter.

Sequence matters: the audit comes before the spend. If you're also re-evaluating paid spend, the calculus there changed less than on organic. Founders running the program themselves can stretch this across two afternoons.

Frequently asked questions

Is SEO dead now that Google has AI Mode?

No. Google's own AEO guide states that "the best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems." Page-level click optimization carries less weight; entity clarity, parsing schema, named expertise, GBP completeness, and crawlability carry more. Listicle content is taking real hits; original work and pages tied to a real business are gaining value.

What does Google's new AEO guide say I should do?

Two lists. Do: indexability and crawlability, JavaScript SEO best practices, semantic HTML, good page experience, less duplicate content, a complete Google Business Profile if you serve a local market, Merchant Center feeds if you sell products. Don't: LLMs.txt files, content chunking, AI-specific rewriting, inauthentic mentions, or special schema beyond what Google already supports. The guide is at developers.google.com/search/docs/fundamentals/ai-optimization-guide.

Should I create an LLMs.txt file?

No. Google's AEO guide places LLMs.txt files at the top of its "you do not need to do" list. Other AI providers may behave differently, but for Google's surfaces the answer is no.

How is AI Mode different from AI Overviews?

AI Overviews are the AI-generated summary block at the top of the ten-blue-links results page for some queries. AI Mode is a separate surface, a full conversational interface where you type in plain English, get a synthesized answer, ask follow-ups, and now use agentic features like 24/7 Information Agents and booking. Both run on Gemini, both can cite your pages, both can earn zero clicks. AI Mode is the future Google is positioning; AI Overviews are the bridge.

Will Google AI Mode have ads?

Not yet, but ad-supported AI looks likely. Google Search and related advertising was roughly 57% of Alphabet's revenue in fiscal 2024 (56.6%, the most recent full-year figure); the company is unlikely to leave a billion monthly AI Mode users unmonetized for long. Google has signaled testing of ad-like placements in AI Overviews, and AI Mode units are widely expected to follow. For a 12-to-24-month roadmap, assume a paid-and-organic hybrid by 2027. I wouldn't bet on the exact timing.

If your pages are getting summarized but not cited, that's the gap to close this quarter. Run your top thirty URLs through our AI visibility checker (no login), then run the free SEO audit to catch the schema and crawlability issues that decide whether the assembler reaches for your page or a competitor's.

Our powerful suite of automation tools for SEO

Learn, discover, and get inspired by our content