The proposal landed in my inbox forwarded by a client with three words: "is this real?" Fourteen pages. New discipline. "Agentic SEO." Separate retainer, separate framework for the thing Google announced three weeks earlier at I/O. The only concrete recommendation I could identify in the whole document was to create an llms.txt file and break their content into smaller chunks for AI readability.
Google had said, explicitly and in published documentation, that neither of those things is required or useful for its generative AI features.
I have watched this pattern enough times to see it coming. A major platform change happens. The industry names it, brands it, and attaches a new service tier within ten business days. The businesses on the receiving end are almost always the same ones who were sold voice search optimization in 2019. Different invoice, same underlying confusion about what actually changed and why it changed.
Here is what Google's Information Agents actually are: autonomous AI systems that operate in the background twenty-four hours a day, reasoning across information including blogs, news sites, social posts, and real-time data on finance, shopping, and sports, to find what a user needs and send them a synthesized update. A user sets an instruction once: "alert me when a supplier in my category drops below $40 per unit" or "track when a property in this neighborhood hits my budget." The agent runs that instruction continuously. Search stops being a session and becomes a standing instruction.
That is a real structural shift. And almost every piece of coverage I read about it focused on the wrong half.
Image credit: Screenshot from "Everything that happened in Google I/O 2026: The New Frontier of SEO and AI Search" by Passionfruit on YouTube (https://www.youtube.com/watch?v=qI6Gt2uM66Q).
The part that matters is not what the agents do. It is when they do it. They are not waiting for a user's query to evaluate your page. They evaluate content on their own schedule, against criteria a user set in the past, to determine whether your site contains a data point worth surfacing in a synthesized update. Your page does not need to rank at the moment of a query. It needs to be accurate, crawlable, and specific enough that an AI system can extract a meaningful fact from it without being asked.
That is a technical SEO problem with a technical SEO solution: correct schema, product information that is actually current, page speed that lets Googlebot complete a crawl pass efficiently. The things that have always mattered for crawl infrastructure suddenly matter for a different reason. A background agent hitting your product page and finding stale pricing is not a ranking event. It is an exclusion event.
Here is where the community got the story backwards. The framing that took hold after I/O 2026 was that businesses need to optimize for AI agents as a distinct discipline, separate from traditional SEO. GEO. AEO. Agent-first content architecture. New retainers, new vocabulary. Google itself stated in published documentation: "From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO." The documentation explicitly named llms.txt, content chunking, AI-specific rewriting, and special schema as things site owners can ignore for Google's generative AI features. That guide came out before the I/O keynote. The industry largely ignored it.
The conventional wisdom about agent optimization broke down in a specific, testable place. Multiple blogs and consultants recommended breaking content into short, discrete chunks to improve AI extractability. Google engineers, when asked directly, recommended against it: Google's systems can understand the nuance of multiple topics on a page and show the relevant piece to users. Fragmented content that strips depth and removes connective tissue performs worse, not better. A page structured like a bare fact table is not more machine-readable. It is thinner, and it signals less genuine expertise to the same systems now powering agent evaluation.
What the agents actually reward is what the May 2026 core update rewarded: content that reflects genuine expertise and says something specific rather than something broad. The agent needs extractable facts to match against a user's standing instruction. A product page that says "competitive pricing" gives the agent nothing. A page that says "$47 per unit, ships same day, available in four variants, compatible with X and Y" gives it something to work with.
Position one click-through rates have dropped from 27 percent to 11 percent. Sixty percent of all Google queries now result in zero clicks. Those numbers are real and they matter. But the response to them is not a new framework. It is better execution of the old one, because the agents reading your content are doing so with higher precision than any previous crawl system, and they surface the gap between what your page claims and what it actually contains faster than any human reviewer would.
If a Search Agent cannot pull one accurate, specific fact from your most important page within a single crawl pass, no file you place in your root directory will change that outcome. That is the whole problem, and it has had a known solution since before anyone announced these agents existed.
Waleed Qamar holds a BSc in Computer Science from Purdue University and has spent the years since turning that technical foundation into something the curriculum never covered: figuring out why websites rank, why they fall, and why most businesses never find out until it is too late.
Pakistan-born and based between the United States and South Asia, he has managed search visibility for e-commerce stores, local service businesses, and SaaS startups across two continents. He started in SEO when guest posting still worked, survived the Penguin update, and has rebuilt client sites from scratch after algorithm hits more than once.
He has watched good businesses get sold packages that looked like progress and delivered nothing lasting. He has also seen the right approach quietly double a site’s traffic without a single press release about it.
His writing on SEO By Highsoftware99 covers Google algorithm updates, autocomplete optimization, semantic SEO structure, and the widening gap between what agencies promise and what Google actually rewards in 2026.
He knows what a traffic cliff looks like in Search Console on the morning you discover it.
