Rankings Won’t Matter If AI Agents Don’t Know Your Brand Exists
Google CEO Sundar Pichai sat down with Stripe co-founder John Collison and investor Elad Gil on the Cheeky Pint podcast earlier this month. The conversation ran over an hour and covered Google’s AI history, infrastructure spending, compute bottlenecks, and where Search is headed. One theme ran through everything Pichai described: Search is moving from returning results to executing tasks, and in that model, the question isn’t where you rank. The question is whether the AI agent knows your brand exists at all.
Pichai used the phrase “agent manager” to describe what Search is becoming. Not a search engine. Not an answer engine. A system that coordinates tasks on a user’s behalf, running multiple threads simultaneously. In that world, the brands that maintain visibility are the ones AI systems already associate with specific topics, services, and areas of expertise. That association, entity recognition, is becoming the most important asset in organic search strategy. Understanding why requires looking at what Pichai actually said and what it means for how discovery works going forward.
What Pichai Actually Said About the Future of Search
When asked what the future of Search looks like, whether it remains a product, a distribution mechanism, or one of many ways people interact with the world, Pichai described something fundamentally different from the search engine that exists today.
He said that information-seeking queries, the kind of searches that currently return a list of ranked results, will become agentic. Instead of typing a question and getting links, users will initiate tasks. Instead of searching for trip options, Search will plan the trip. Instead of returning results, Search will coordinate actions across multiple services simultaneously.
Pichai used the phrase “agent manager” to describe what Search becomes in this model. Not a search engine. Not an answer engine. A system that manages concurrent tasks on a user’s behalf, with multiple threads running at once. He compared it to how he already uses internal Google tools, where agents perform work in parallel and the user oversees the output rather than doing each step manually.
When asked whether the traditional search box will still exist in ten years, Pichai avoided a direct answer but said something revealing. He said the form factor of devices will change, that input and output methods will change radically, and that trying to predict ten years out is paralyzing. Instead, he said the curve of AI progress is so steep that thinking one year ahead is more productive, and more exciting, than trying to envision five years out.
He also addressed the relationship between Search and Gemini, Google’s standalone AI model. He said the two will overlap in some areas and profoundly diverge in others, and that Google is committed to running both. AI Mode in Search serves as the bleeding edge, and features that prove successful there migrate to the main search experience over time. The implication is that AI Overviews and AI Mode are not experiments. They are staging grounds for the next version of Search.
What “Agent Manager” Means in Practice
To understand what Pichai is describing, it helps to think about how Search works today and what changes when it becomes agentic.
Today, a user types a query. Google returns a ranked list of results, increasingly topped by an AI Overview that summarizes information from multiple sources. The user clicks a result or reads the overview. The interaction is essentially one question, one response, one session.
In an agentic model, the interaction looks completely different. A user describes what they want to accomplish rather than what they want to know. Search doesn’t return results. It initiates a process. It might contact multiple services, compare options, negotiate parameters, and execute a decision, all within the search interface, all without the user visiting a single external website.
Pichai referenced 2027 as a potential inflection point for when agentic workflows expand beyond engineering and developer use cases into mainstream commercial applications. That timeline aligns with what most infrastructure observers expect for broad deployment of long-horizon AI agents.
The practical example is straightforward. Today, someone searching for a weekend trip to Lisbon gets a list of blog posts, hotel booking sites, and flight comparison tools. In an agentic model, Search takes the intent, the budget, the dates, the user’s preferences, and books the trip. The blog posts, the hotel sites, and the comparison tools are still data sources, but the user never visits them. Search consumed their information and acted on it.
Why This Matters for Organic Search Strategy
The entire architecture of SEO is built on a foundational assumption: that Google Search exists to connect users with websites. Every ranking factor, every backlink, every piece of optimized content operates within that framework. Users search. Google returns links. Users click. Websites get traffic.
Pichai’s description of the future doesn’t include that loop.
That doesn’t mean websites stop existing or that SEO becomes irrelevant overnight. Google still needs sources. AI Overviews and agentic processes still need data to synthesize, services to connect to, and information to act on. But the role of a website in that ecosystem changes from a destination to a data source. The user doesn’t visit you. The agent consumes your content and uses it as an input.
For link building and digital PR, the implications ripple outward. Backlinks have always been a signal of authority and trust. When Search operated as a referral mechanism, links mattered because they influenced which results users saw and clicked. In an agentic model, authority signals still matter, possibly more than ever, because the agent needs to decide which sources to trust when executing tasks. But the downstream value of the link changes. A backlink that used to drive referral traffic and pass ranking authority might still pass authority, but the referral traffic component diminishes if the agent is doing the visiting on the user’s behalf.
Guest posting and contributed content face a similar recalibration. Publishing an article on a respected industry site has traditionally served dual purposes: earning a link for SEO value and putting a brand in front of that publication’s audience. If the audience increasingly gets their answers through agentic search without visiting the publication, the second purpose weakens. The SEO value of the link may persist, but the brand exposure depends on whether the agent surfaces the brand name in its output, which is a question of entity recognition, not traditional ranking.
Entity Recognition Becomes the Core Asset
In a search environment where agents synthesize information and execute tasks without sending users to websites, the brands that maintain visibility are the ones the agent knows about and trusts. That knowledge comes from entity recognition, the ability of AI systems to associate a brand name with specific topics, services, and areas of expertise.
Entity recognition gets built through exactly the kind of work that has always supported strong SEO: consistent publishing on authoritative sites, editorial coverage that mentions the brand in context, topical relevance reinforced across multiple sources, and structured data that helps AI systems understand what a company does and where it operates.
The difference is that in a traditional search model, all of that work eventually translated into rankings, and rankings translated into clicks. In an agentic model, the translation step changes. The work still matters, but its output is whether the agent includes your brand when executing a task, not whether you appear on page one of a results list.
Companies that have invested in building strong entity association through consistent link insertion, editorial mentions, and topical authority across the web are better positioned for an agentic search environment than companies that have optimized primarily for keyword rankings. Keyword rankings assume a results page. Entity recognition works regardless of the interface.
What Pichai Didn’t Address
Pichai’s interview was notable for what it left out. He didn’t discuss how publishers will be compensated when their content is consumed by agents without generating traffic. He didn’t address how advertising works in an agentic model where users don’t see search results pages. He didn’t explain how smaller businesses, the ones that don’t have the brand recognition to be known to AI agents, will get discovered in a world without clickable results.
These are not minor gaps. Google’s advertising business generated over $113 billion in a single quarter in late last year. That business depends on users seeing and clicking ads in search results. An agentic model that executes tasks without showing results pages disrupts the revenue model that funds everything Google does.
Pichai did say that commercial information will still have value, and that AI will help figure out the best way to integrate it. But the specifics remain undefined. For businesses that depend on organic search traffic for revenue, the absence of a clear answer about how discovery and monetization work in an agentic model is the most important thing Pichai didn’t say.
The Timeline Question
How quickly this transition happens is the most practical question for anyone making SEO and content investments today.
Pichai’s comments suggest the transition is already underway. AI Overviews are live globally. AI Mode is available as a separate tab for users who want the more advanced experience, with successful features migrating to the main search page over time. The underlying models are improving on a curve steep enough that Pichai said thinking more than a year ahead is less productive than just riding the current trajectory.
At the same time, Pichai acknowledged that the full agentic vision is not here yet. He referenced 2027 as a potential inflection point for non-engineering use cases. The infrastructure constraints are real. Google is spending $175 to $185 billion in capital expenditures in 2026 and still can’t build fast enough because memory chips, wafer capacity, and power infrastructure are the binding constraints, not money.
That means there’s a window, but the window has a visible end. The traditional search model isn’t going to disappear in the next twelve months, but the signals about where it’s heading are no longer ambiguous. The CEO of Google described the future of search as a task execution system and didn’t mention websites. That’s about as clear a signal as you can get.
What to Do With This Information
The practical response is not to abandon SEO or stop building links. The foundational work of earning authority, building brand recognition, and establishing topical relevance across the web is the same work that positions a brand for visibility in an agentic search environment.
The adjustment is in how you measure success and where you direct effort. Traffic from organic search will likely decline over time as agentic features absorb more query types. But brand mentions in AI generated outputs, entity recognition across AI systems, and the authority signals that come from appearing on respected third party sites will become more important, not less.
Building a strong backlink profile through quality editorial placements is still valuable, and may become more valuable, because those placements contribute to the entity recognition that determines whether an agent includes your brand when executing a task. The measurement just needs to expand beyond rankings and traffic to include brand visibility in AI outputs.
Pichai gave a one-year planning horizon as the most productive timeframe for decision-making. That’s a reasonable framework. Invest in the foundational work that serves both the current search model and the emerging agentic model, and reassess as the product evolves. The worst position to be in is having built nothing that AI systems recognize when the transition completes.