People
They read your page and decide whether to contact you.
GEO · Generative Engine Optimization
More and more searches are carried out by agents.
A specialized service so your site is found where language models look for their answers.
Companies we have worked with
What your report doesn't show you
If AI can't find your site, it doesn't matter how hard you've worked to get Google to show it in its results.
Your customer doesn't always open Google anymore. They ask ChatGPT, Gemini, Perplexity, or a corporate assistant who the best provider of what you sell is. The answer arrives already written, with the sources the model chose.
Your site can be in first position on Google and not be in any of them. Your SEO reports won't tell you: they don't measure this.
Position on Google1
Mentions in the AI's answer0
Illustration of the case
How an answer is built
If AI doesn't cite you, the answer is built without you.
A generative engine doesn't show you a list of ten links. It extracts fragments from public sources, processes them with a language model, and synthesizes an answer. Your site doesn't appear in a results list: it is cited as a source or it doesn't exist in that answer.
And it isn't just your customer's ChatGPT. Companies are deploying internal assistants on top of models from OpenAI, Anthropic, Google, or open models on their own infrastructure, and each one searches the web for information and decides which sources to cite. If your site isn't built to be read by them, you don't exist in those answers.
Cited as a source, or absent from the answer.
The third reader
The site you built for people and optimized to show up on Google now has a third reader.
Search engines
They index your pages and rank them in a results list.
AI agents
They read a visual capture, structural HTML, and the accessibility tree, and decide whether to cite you in their answer.
AI agents read your site in three ways at once: visual capture, structural HTML, and accessibility tree. If the HTML isn't semantic, if ARIA attributes are missing, if the hierarchy is ambiguous, or if the content depends on JavaScript to render, the model doesn't process it well, and a company with years of SEO can be completely absent from those answers without knowing it.
The decision isn't how much content to publish. It's whether your site is built to be read by a machine.
What we do
We make sure AI agents mention your brand in the right context. Control the message.
Visibility audit
We measure whether ChatGPT, Gemini, Perplexity, DeepSeek, Copilot, and Google AI Overview cite you today, and who they cite instead.
Diagnosis and implementation
We review how readable your site is to a model: semantic HTML, ARIA attributes, accessibility tree, Core Web Vitals, and content crawlability.
We implement structured data (Schema.org) and the technical markup that models interpret, and we bring your brand's consistency as an entity into order on the platforms where models look for context.
Becoming a cited source
We define the content strategy to make you a cited source, monitor appearances and mentions, and train your content team to produce under these criteria.
No implementation moves forward without the baseline. Every month we compare against it and record which change produced which effect.
Proof
We start with our own site.
Readable by language models
theleanmarketing.com is built with semantic HTML, structured data, and an llms.txt file with the company's key information.
See llms.txtOpen to AI crawlers
Our robots.txt explicitly allows the crawlers of OpenAI, Anthropic, Perplexity, Google, and others.
See robots.txtQuestions and answers as data
Our frequently asked questions are marked up as structured data so a model can extract them as an answer.
See the questionsBenefits
What changes when your site is built to be cited.
Presence where the purchase is decided
We lay all the groundwork so language models find your content, read it, and share it as part of their answers.
Your SEO isn't wasted
GEO doesn't replace SEO: it complements it. They share technical foundations, and GEO adds the visibility layer in the generative AI ecosystem.
You know what worked and why
Periodic audits across the main generative engines, compared against an initial baseline: appearance, quality of the mention, and position as a cited source.
Frequently asked questions
Before we talk, this is what people usually ask.
Another question? Write to us and we'll answer it.
GEO is the discipline of optimizing a site so language models — ChatGPT, Gemini, Perplexity, DeepSeek, Ernie, Qwen, Copilot, and enterprise agents — find it and show its content as a source when building answers. It is the evolution of SEO toward the generative AI ecosystem.
SEO optimizes so Google indexes and ranks pages. GEO optimizes so LLMs select your site as a source via RAG. Models prioritize topical authority, semantic HTML, and entity consistency over keyword density.
Yes. It is more common than it seems. A site in position 1 on Google can be completely absent from the answers generated by ChatGPT or Perplexity because the logic for selecting sources is different.
We cannot guarantee that you will be found by these services. In fact, not even their own engineers can guarantee it. What we do is lay all the groundwork so they can find the content easily, read it, and share it as part of their answers. We rely on the guidelines published mainly by Google on this subject, and we constantly evaluate the content published by ChatGPT, Gemini, Perplexity, DeepSeek, Ernie (Baidu), Qwen (Alibaba), Microsoft Copilot, Google AI Overview, and some internal enterprise agents that use LLM APIs on their own infrastructure.
Correct semantic HTML, ARIA attributes, a readable accessibility tree, Schema.org implemented, a robots.txt without blocks on AI crawlers, content that renders without JavaScript, and an llms.txt with the company's key information.
RAG (Retrieval Augmented Generation) is the mechanism LLMs use to look up external information before answering. Your site doesn't "appear in results": it is extracted as a piece of knowledge or it doesn't exist in that answer.
Technical changes take effect in weeks. Changes in content and topical authority take between 2 and 4 months to be reflected in visibility within generative engines. Once again: we cannot guarantee that you will be found or how well you will rank.
It doesn't replace it, it complements it. SEO is still necessary for visibility in traditional search. GEO adds the visibility layer in the generative AI ecosystem. Both share technical foundations but require additional specific work. SEO so humans find you, GEO so AI agents find you.
Visibility audit across generative engines, technical diagnosis, implementation of changes, content strategy for RAG, brand entity consistency optimization, llms.txt, and training for the content team.
With periodic audits across the main generative engines, compared against the initial baseline. We measure appearance, quality of the mention, and position as a cited source.
Next step
Find out whether language models already cite you.
A human-to-human conversation about how your site appears in AI answers.
No commitment.