SEO for AI search: get found on ChatGPT and Google

Sometimes an enquiry arrives saying “ChatGPT recommended you”, and Google's results page increasingly opens with an AI summary. SEO is not dead in the age of AI search, but it has changed.

Why it matters in 2026

It is no longer only about your position in the results list. It is also about whether Google AI Overviews or a ChatGPT answer pulls you in as a source. If you ignore this in 2026, you miss a channel that is quietly taking over part of the role of classic search.

Search is no longer ten blue links

For years, a Google results page meant choosing from ten links. Today you increasingly get an AI summary at the top of the list, and it answers many questions without a single click. Some users never reach Google at all.

They ask in ChatGPT search, Perplexity or Copilot, in full sentences, not keywords: “Which Hungarian company builds AI customer service for SMEs?” The answer to that is not a results list but two or three concrete recommendations.

What the shift means

This shift means two things. First, some clicks disappear, because the machine gives the answer itself. Second, anyone who makes it into the AI answers gets a disproportionate share of attention.

You do not appear next to ten competitors, but next to two or three, or alone.

AEO and GEO: what the new acronyms mean

Two abbreviations are spreading: AEO (answer engine optimization) and GEO (generative engine optimization). AEO is about structuring your content so that it gives a short, quotable answer to a specific question straight away.

GEO aims to make generative models, such as ChatGPT, Gemini and Perplexity, cite you as a source when someone asks about your topic.

The good news is that neither replaces classic SEO. Both build on it. A fast, technically sound, easy-to-crawl site, a clear information structure and real expertise are the foundation, as before.

The difference is in the emphasis. Clarity for the machine and a question-and-answer logic matter much more now.

Structured data: how a machine knows who you are

Structured data (schema.org markup) gives machines a readable description of the company and content, so search engines guess less. Google says it is not required for its AI features. Organization markup says who the company is and what it does. Article says who wrote a piece and when.

FAQPage marks up question-and-answer pairs in machine-readable form. Since 2023 Google shows FAQ rich results only for well-known government and health sites. Service and LocalBusiness markup make clear what service you provide and where.

This is not magic. It is groundwork, and that is why it pays off. You do it properly once, and every later piece of content benefits.

In every web development project we treat it as standard equipment, not a later extra.

llms.txt: the little sibling of robots.txt

llms.txt is an unofficial format proposed in 2024: a simple text file in the root of your domain that tells AI systems what your site is about and where to find the most important content. To be honest, Google says Search ignores it, and no major AI system is known to demonstrably use it, so it does not bring traffic on its own.

It is quick to make and does no harm, but it is a side step: the substance is good content and a sound technical base.

Entity-based SEO: the age of keyword stuffing is over

Modern search engines and language models do not count keywords. They think in entities: companies, people, services, places, and the relationships between them. The goal is a clear picture of you in the machine's knowledge graph: this company does this, here, for this long, and is credible on these topics.

In practice this means three things:

• Consistency: your company name, activity description and contact details say the same thing on every platform, including your own site, company registries and social profiles.

Connections and mentions

• Connections: your content links to itself and to your services, so it is clear which topics belong to you.
• Mentions: every outside mention, whether a trade article, an interview or a guest post, strengthens the entity, even when it is not a classic link.

Why real expert content is worth more

There is an uncomfortable truth here. Generic “5 tips” content that anyone can write can now be produced by AI itself, so its value is quickly heading to zero.

What a machine cannot produce is first-hand experience: how a project runs at your company, what tends to go wrong and what you learned from it, and what questions customers really ask.

Google's quality guidelines (E-E-A-T: experience, expertise, authoritativeness, trustworthiness) and its guidance for AI features both favour unique, people-first content.

Content published under a real name, with a face and a real professional background, is no longer an extra. It is the entry ticket.

What to do now: a practical order

1. Technical base: fast loading, clean HTML structure, a mobile-friendly site, crawlable content.
2. Structured data: Organization, Service, Article and FAQPage markup on your key pages.
3. Question-and-answer blocks: real customer questions with short answers on every service page and article.
4. llms.txt: optional, quick to do.
5. Entity clean-up: consistent company details and descriptions on every platform.
6. Expert content at a steady pace: less, but deeper. It is also worth turning production and distribution into a system, as we do in our marketing automation work.
7. Measurement: track traffic from AI sources separately, for example visitors referred from chatgpt.com or perplexity.ai, so you can see what works.

What it brings, and when

To be honest, optimising for AI search is not a one-week project, and it is not a guaranteed machine. As a rough guide, the basics (structured data, llms.txt, entity clean-up) can be put in place within a few weeks.

A meaningful effect is usually measured in months, as search engines and models re-index and re-evaluate the site.

Where to start

We build both websites and AI systems, so we see both sides of how a machine reads content. If you want to know where your site stands, we will review it: technical state, structured data, entity picture and content.

A focused project starts from a few hundred thousand HUF, and the basics can often be fixed for that. Write to us and we will tell you honestly what you need, and what you do not.