Library room with wooden cabinets full of card catalog drawers
Guides

llms.txt and JSON-LD: how to get AI to recommend your website

ChatGPT, Gemini, and Google no longer just link to pages: they read your site and answer on your behalf. We explain, in plain words, what llms.txt and JSON-LD are, why they matter for SEO and the new GEO, and how they apply to your business.

Damorus Projects9 min read

Not long ago, getting found online meant "showing up on Google." Today, more and more people ask an AI directly: "who can build me a website in Bogotá?" or "recommend a good, affordable web design agency near me." And the AI answers with a short list of businesses, without the person ever opening ten tabs. The million-dollar question is: why would it recommend you and not your competitor?

The answer has two protagonists with technical-sounding names that are actually simple: llms.txt and JSON-LD (also called structured data or schema.org). Neither one shows up on your page; both speak to machines. In this guide we explain them the way we would over coffee with a business owner, without unnecessary jargon.

LO QUE VE EL HUMANOLO QUE LEE LA MÁQUINATu negocioBogotáDesde $CONTACTARnameaddresspricecontactllms.txt
Your page looks the same to the customer. Underneath, two invisible layers speak to the machines.

First, two new acronyms: SEO and GEO

SEO (Search Engine Optimization) is the classic one: preparing your site so search engines like Google understand it and show it when someone searches. GEO (Generative Engine Optimization) is the new version: preparing your site so generative AI engines, like ChatGPT, Gemini, Copilot, Perplexity, and Google's own AI summaries, understand you and recommend you inside their answers.

IA · RESPUESTA10 enlaces3 recomendados{ }
A search engine gives you ten links. AI pulls out two or three recommendations: either you are in the answer, or you do not exist.

What is llms.txt? Your site's "index for robots"

Think of walking into a huge shopping mall: the first thing you look for is the map with the "you are here" dot. llms.txt is exactly that, but for AI. It is a very simple text file that lives at the root of your domain (it sits at yourdomain.com/llms.txt) and tells language models: "here is what matters on this site, and here is where everything lives." [1]

The proposal was born in 2024 from Jeremy Howard (co-founder of Answer.AI) and has been gaining ground as an open standard. [1] [2] The problem it solves is real: modern web pages are full of menus, buttons, banners, and code that are hard for an AI to digest. llms.txt hands over a tidy version, in Markdown format (plain text with headings and links), so the machine does not get lost and understands your business on the first try. [2]

This is how simple an llms.txt file looks on the inside. Notice that it is readable even for you: it is just a title, a description, and a list of links with short explanations.

markdown
# Damorus Projects

> Web design and development studio. Professional, fairly priced sites
> for small businesses, with public "Starting at" pricing.

## Plans
- [Silver Plan: digital launch](https://damorus.com/en/plans/plata): a fast landing page starting at USD 300.
- [Gold Plan: digital authority](https://damorus.com/en/plans/oro): a custom multi-page site starting at USD 1,000.
- [Diamond Plan: unlimited](https://damorus.com/en/plans/diamante): a premium e-commerce build starting at USD 5,000.

## Resources
- [Request a quote](https://damorus.com/en/start): a plan recommendation based on your goal and budget.
- [Blog](https://damorus.com/en/blog): guides on SEO, GEO, and web design.
Example of /llms.txt for a studio like Damorus Projects.

Let's be honest: llms.txt is a young proposal, not every AI reads it the same way yet, and on its own it does not guarantee you will "rank." [2] But it costs very little to set up and points in the right direction: giving machines a clean summary of who you are. Worst case, it does no harm; best case, it makes you the easy option to cite.

What is JSON-LD (schema.org)? The "spec sheet" machines actually understand

You read your page and instantly understand that "Damorus Projects" is a company, that "COP 1,000,000" is a price, and that "Bogotá" is a city. A machine, on the other hand, only sees loose text: it does not know what is a price, an opening hour, or a phone number. JSON-LD is the way of tagging that information so the machine understands it without guessing.

The tagging vocabulary is called schema.org, and it is not just anyone's invention: it has been jointly maintained by Google, Microsoft (Bing), Yahoo, and Yandex since 2011. [4] In other words, it is the common language the major search engines agreed on to describe things in the real world (a company, a product, an event, a recipe, a review...). JSON-LD is simply the recommended format for writing that language inside your site. [5]

Damorus Projects$300Austin???etiquetadoranamepricecity?✓nameDamorus Projectsprice$300cityAustin
The machine only sees loose text until someone tags it: name, price, city.

It goes inside an invisible block within the page's code (a <script type="application/ld+json"> tag). It changes nothing your customer sees: it is a separate message, meant only for search engines and AI. Google recommends it over other formats precisely because it is clean and does not clutter the design. [5] [7]

json
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "ProfessionalService",
  "name": "Damorus Projects",
  "description": "Engineering, science, and AI studio that builds complete digital projects.",
  "url": "https://damorus.com/en",
  "telephone": "+49-176-4454-2881",
  "priceRange": "$500 - $5,000 USD",
  "areaServed": [
    { "@type": "City", "name": "Bogotá" },
    { "@type": "City", "name": "Berlin" },
    { "@type": "Country", "name": "Colombia" }
  ],
  "makesOffer": [
    {
      "@type": "Offer",
      "name": "Silver Plan",
      "priceCurrency": "USD",
      "price": "500"
    }
  ]
}
</script>
Real JSON-LD from Damorus Projects: it describes the business, the cities it serves, and its plans as "offers."

Read it without fear: "@type": "ProfessionalService" means "I am a professional service"; areaServed lists the cities where you work; and each Offer is a plan with its price. When Google or an AI reads this, it stops guessing: it knows exactly what you offer, where, and starting from how much. That is the whole trick.

Why this matters to you (even if you are not technical)

  • You show up "enriched" on Google. Structured data is what triggers rich results: review stars, prices, expandable FAQs. That gets more people to click on you. [5] [7]
  • AI can cite you. A business with tagged information is an easy business for ChatGPT, Gemini, or Google's AI summaries to recommend. You hand them the answer, ready to go. [8]
  • You reduce misunderstandings. If your price, phone number, or hours are tagged, it is far less likely a machine will show outdated or wrong information about your business.
  • It is cheap and lasts. You set it up once and it works for you 24/7, with no paid ads. It is one of the best effort-to-reward ratios in digital marketing.

“JSON-LD is the structured-data format Google prefers. It helps search engines understand your content and can improve how you appear in results.”

Yoast SEO, summarized from its guide on JSON-LD

How does this work in practice?

DE TU INFORMACIÓN A LA RESPUESTAte recomiendo:tu negocio ✓ORGANIZAETIQUETARESUMEVALIDAMANTIENEcada paso empuja al siguiente
From organizing your information to getting cited by an AI: each step pushes the next one forward.
  1. 1We organize the information. We define the essentials: what you do, where, "starting at" prices, phone number, hours, and real frequently asked questions.
  2. 2We tag it with JSON-LD. We translate that into schema.org (business, offers, FAQ, reviews) and insert it as an invisible block in your site. [4] [6]
  3. 3We summarize it in llms.txt. We publish an llms.txt with the key links, so AI has a clean map of your site. [1]
  4. 4We validate it. We use Google's own tools to confirm the data reads without errors. [5]
  5. 5We maintain it. When a price or a service changes, we update the tags so machines always show the right information.

In short: llms.txt is the map you hand AI, and JSON-LD is the spec sheet you hand search engines. Neither one is visible, both work in your favor, and together they make sure that when someone asks about a business like yours, the name that comes up is yours. Want your site ready for this new era? Request a quote and we will get it running.

Sources

References cited in the text with the [n] notation. Each link was verified and is publicly accessible.

  1. 1

    El archivo /llms.txt: documentación oficial de llms.txt (Jeremy Howard, Answer.AI y la comunidad abierta)

    Official specification: defines what llms.txt is, its Markdown structure, and how to implement it to optimize content for language models. Primary source.

    https://llmstxt.org/
  2. 2

    llms.txt, un estándar propuesto para el rastreo de contenido web por IA (Search Engine Land)

    Accessible analysis of the proposed standard, its purpose for AI-oriented optimization, and its place within the modern search ecosystem.

    https://searchengineland.com/llms-txt-proposed-standard-453676
  3. 3

    llms.txt: qué hace y cómo encaja en la optimización para búsqueda con IA (Stellar AEO Labs)

    Practical explanation of how llms.txt supports RAG (retrieval-augmented generation) operations and improves the accuracy of LLM answers.

    https://stellar-ai.co/blog/llms-txt-ai-search-optimization/
  4. 4

    Schema.org: sitio oficial de la comunidad (Google, Microsoft, Yahoo, Yandex y la comunidad abierta)

    Primary source for the structured-data vocabulary: defines what schema.org is, who maintains it, and documents its types and properties. Primary source.

    https://schema.org/
  5. 5

    Introducción a cómo funciona el marcado de datos estructurados (Google Search Central)

    Official Google documentation on structured data: benefits, supported formats (JSON-LD, Microdata, RDFa), and best practices. Primary source.

    https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data
  6. 6

    Primeros pasos con schema.org usando Microdata (Schema.org)

    Primary technical tutorial on how to code structured data, with practical schema.org implementation examples.

    https://schema.org/docs/gs.html
  7. 7

    Usa JSON-LD para agregar datos de schema.org a tu sitio web (Yoast SEO)

    Practical guide on why JSON-LD is Google's preferred format and its benefits for CTR and rich snippets.

    https://yoast.com/json-ld/
  8. 8

    Principales patrones de schema JSON-LD para SEO que impulsan la visibilidad en búsqueda con IA (Growth Natives)

    Analysis of JSON-LD patterns that improve visibility in AI-driven search, including their impact on Google AI Overviews and similar systems.

    https://growthnatives.com/blogs/seo/top-json-ld-schema-patterns-for-ai-search-success/

Know someone this helps? Share it.

Photo: “Sterling Memorial Library card catalog” by pushypenguin, CC BY 2.0, via Wikimedia Commons.

Your project with us can start today.

Tell us what you want to achieve, or that thing you always wanted your project to achieve, and we will tell you how we would do it and what it costs. If you can imagine it, we can make it real.

or email us at Alejandro@damorus.com