Blog// artificial intelligence

How the Agent Layer Is Consuming Search and What Businesses in Russia and the CIS Should Do

How AI agents are replacing search and what businesses in Russia should do: the shift from SEO to GEO, structured data, and a focus on real customer tasks.

22-10-2026 · 8 min read
How the Agent Layer Is Consuming Search and What Businesses in Russia and the CIS Should Do

Google CEO Sundar Pichai has acknowledged that traditional link-based search is fading into the past, giving way to the agent layer. AI agents themselves perform user tasks — booking, comparing, buying — without requiring a person to click through results. As Habr writes, for a company whose business model is built on clicks, this statement resembles either rare honesty or expensive investor expectation management.

The shift in approach strikes at the small business sales funnel. Previously, the marketer's task was to land in the top ten links. Now the question is different: how to get into the context of an agent that assembles an answer and acts on the client's behalf. If your product or content does not address a specific human task, optimizing headlines and meta* tags will not help. The agent operates as a filter for substance, not for algorithmic compliance.

From Link Listings to Task Execution

The essence of the shift Pichai describes is simple: the user sets a goal, not a query. “Book a table for two downtown on Friday evening” — and the agent searches for available spots, checks reviews, clarifies the menu, books, and adds the event to the calendar. The person does not see a results page, does not go to the restaurant's website, does not read a map listing.

In this model, Google Search does not disappear but becomes an internal system. It supplies data to the agent: schedules, prices, product availability, ratings, page content. But the interface — a dialogue with the agent — hides this layer. For business, this means losing direct contact with the customer at the search stage. The entry point shifts from your website or service listing to the agent's answer.

According to Habr, Pichai calls this a “reimagining” of search. Outwardly it looks like an evolution of convenience. Inside, it is a redistribution of traffic and attention. Whoever controls the agent controls access to the customer.

Why Google Is Saying This Now

The statement is not accidental. ChatGPT, Perplexity, and other AI tools are already capturing scenarios that used to start with the Google search bar. Users are getting used to asking complex questions and receiving ready-made, detailed answers without clicking links. The share of such sessions is growing, while monetization of the classic results page is under threat.

The phrasing “search will absorb the agent layer” sounds better than “we are losing control over search.” This is rhetoric for shareholders: we are not retreating, we are leading the process. But the reality Habr notes is simple: agents are changing how a person reaches a result. Google is forced to build its own agent products — AI Mode, AI Overviews — to keep users in its environment.

For business, it is important to separate the marketing wrapper from the mechanics. Agents are already here. They scan the web, use structured data, read reviews, analyze catalogs. If your information is not available in machine-readable form — it does not exist for the agent.

The New Logic of Visibility: How to Get Into the Agent's Context

Old SEO answered the question: how to be visible to the ranking algorithm. The new one — GEO (Generative Engine Optimization) — answers the question: how to be useful to the agent that makes decisions on behalf of a person.

The agent does not evaluate website design and does not count keyword density. It looks for verified facts: prices, specifications, delivery terms, reputation, current stock. It assembles an answer from multiple sources and cites those it trusts. Trust is built from domain authority, data freshness, structured markup (Schema.org), reviews from real people, and mentions in authoritative directories.

The principle older than any update remains in force: the need comes first. Marketing never created demand — it followed existing scarcity. The agent does not cancel this mechanic; it exposes it. It arrives to close a specific task. If your product or your text closes that task — you will be included in the answer. If not — no optimization will save you.

Substance as the Only Filter

In the world of agents, the winner is not the one who shouts loudest in meta* tags, but the one who has something to show. AI operates as a filter for substance. It filters out empty landing pages, doorway pages, content “for robots,” keyword-stuffed texts without specifics.

This is perhaps the most useful thing that could have happened to promotion in recent years. It took the emergence of agents to return marketing to a normal state: make products and content that actually solve a human's task.

Practically, this means a change in the quality criterion. Before: “be visible to the algorithm.” Now: “be useful to the one making the decision for the person.” The difference is enormous. The first breeds imitation of activity. The second requires understanding the client's pain, honest description of product capabilities, accurate data, and ease of action.

What This Means for Small Business in Russia and the CIS

The Russian market has its specifics: Yandex dominates, its own AI models are developing (YandexGPT, GigaChat), different directories and maps (Yandex Maps, 2GIS, Avito, Profi.ru). But the vector is unified — agents are appearing everywhere. Alice is learning to act as an agent, Yandex is integrating neural responses into search, marketplaces are launching AI shopping assistants.

Concrete steps for this week:

  1. Client task audit. Write down 10–15 main goals people come to you with: “buy size 42 gray sneakers with delivery tomorrow,” “book a dentist appointment for the weekend without prepayment,” “select a home pull-up bar under 15 thousand.” Check whether your pages fully address these goals: are there price, availability, terms, photos, reviews, action button.
  2. Structured data. Schema.org markup (Product, Service, LocalBusiness, FAQ, Review) is the language agents speak. Without it, your content is just text to a machine. Check with Yandex and Google validators that the markup is valid and complete.
  3. Reputation and reviews as a data source. Agents trust aggregated ratings and review text. Work with reviews on maps, marketplaces, specialized platforms. Respond, address negative feedback, ask for detailed reviews. This is not about rating for rating's sake — it is about data for the agent.
  4. Catalog and feed currency. If you are in Yandex Business, on Avito, in Yandex Market — feeds must be perfect. Prices, stock, photos, attributes (color, size, material) — all this is read by agents when selecting an option. A feed error means absence from the agent's recommendation.
  5. Content addressing objections. The agent looks for answers to customer fears: “is there a warranty,” “how to return,” “what are delivery times to my city.” Create pages or blocks that directly answer these questions. Not in the website footer in small print, but in structured form on the product or service card.
  6. Local context. For service business and retail, presence in Yandex Maps and 2GIS with a complete profile is critical: interior or storefront photos, menu or price list, operating hours, photos of work or products. Yandex and Alice agents rely on this database.
  7. Do not wait for “new SEO.” Start improving the core of your product and information about it today. Agents do not forgive “we'll do it later.” They index what exists now.

Summary

Search is not dying — it is moving under the hood. The client interface becomes the agent, and business visibility depends on how useful your data is for making decisions on behalf of a person. No algorithm tricks remain. What remains is work on the product, honesty of information, and ease of action. Whoever understands this before competitors will get agent traffic. The rest will wait for clicks that will no longer exist.

Frequently Asked Questions

What is an agent layer in the context of search?

The agent layer is a software system powered by artificial intelligence that interprets user queries, aggregates information from various sources, and provides ready answers or solutions, bypassing direct interaction with traditional search results.

How do AI agents affect SEO?

AI agents shift the focus from traditional SEO, which is keyword- and link-oriented, toward optimizing for solving specific user tasks. This requires content to be more relevant, accurate, and to use structured data.

What is GEO optimization and why is it important?

GEO optimization (Goal-Oriented Optimization) is an approach to creating content aimed at directly meeting user needs and achieving their goals. It is becoming key because AI agents look for not just information, but ready solutions and answers to specific questions.

  • Meta Platforms Inc. activity (Facebook and Instagram* social networks) is recognized as extremist and prohibited on the territory of the Russian Federation.
  • Meta Platforms Inc. (including social networks Facebook and Instagram*) is recognized as an extremist organization and its activities are prohibited in the Russian Federation.

Based on: Habr

Analysis & review by
Alexander K.

Web development, digital advertising and marketing expert since 2012. Founder of AIFORSALE.

How we can help
Online storesMarketingAI agentsContact us
← All articles