The llms.txt file is a Markdown document placed in the root directory of a website (e.g. tuodominio.com/llms.txt), designed to guide Large Language Models such as ChatGPT, Claude and Gemini in accessing and understanding content. The proposal was put forward by Jeremy Howard, co-founder of Answer.AI, in September 2024, in response to a concrete problem: AI models struggle to navigate complex websites full of JavaScript, adverts and visual layouts. llms.txt offers them a curated editorial map.
Who needs to know about it? Anyone managing a digital brand. When did it emerge? In 2024, but its relevance has exploded in 2026 with the growth of AI search. Why does it matter? Because online visibility no longer runs through Google alone.
How It Works: Structure and Variants
The file follows a precise Markdown hierarchy. Alongside llms.txt there is also llms-full.txt, an extended version containing the full text of the main pages.
| Element | Mandatory | Function |
|---|---|---|
| H1 (brand name) | Yes | Identifies the project |
| Blockquote (summary) | No | Contextualises content |
| H2 sections (categories) | No | Organises resources |
| Annotated links | Yes | Guides AI towards key pages |
Copy the LLMS.TXT File Template
# Nome Brand
> Descrizione breve del brand: cosa fa, a chi si rivolge, qual è il suo posizionamento. Max 2-3 righe.
## Pagine principali
- [Home](https://tuodominio.com/): Panoramica del brand e dei servizi.
- [Chi siamo](https://tuodominio.com/chi-siamo/): Storia, mission e team.
- [Servizi](https://tuodominio.com/servizi/): Elenco completo dei servizi.
- [Contatti](https://tuodominio.com/contatti/): Come richiedere una consulenza.
## Contenuti editoriali
- [Blog](https://tuodominio.com/blog/): Articoli su strategia, settore e approfondimenti.
- [Caso studio](https://tuodominio.com/case-study/nome/): Descrizione progetto e risultati.
## Escludi
- /wp-admin/
- /carrello/
- /checkout/</pre>
</div>llms.txt vs robots.txt: A Strategic Distinction
robots.txt controls crawler access. sitemap.xml facilitates indexing. llms.txt, by contrast, is an advisory document with no directive power: an AI model is free to ignore it or follow it. It does not protect content; it prioritises it. It is a tool for editorial governance, not for security.
Why It Matters in 2026: The Market Data
According to Gartner, by 2026 the volume of traditional searches on Google will fall by 25% as a result of AI chatbots. Perplexity processed 780 million queries in May 2025 alone (Source: Index.dev, 2025). In this scenario, being cited correctly by an LLM has become a strategic objective on a par with organic search positioning.
Research by SE Ranking across 300,000 domains found that only 10.13% of websites have an llms.txt file: a real advantage for those who adopt it today.
Real-World Adoption and Case Studies
Brands that have already implemented the standard include Anthropic, Stripe, Cloudflare, Zapier and Cursor (Source: Peec AI, July 2025). Anthropic has published an llms-full.txt with over 480,000 tokens of technical documentation.
In Bliss‘s work, one principle emerges consistently: editorial quality is the prerequisite for any AI visibility strategy. Projects such as Luxury Still Life and the Guida per Elevare un Brand show that structured, authoritative, hierarchically organised content is exactly the material llms.txt is able to bring to the fore, offering models a direct path to the brand’s positioning. No technical file makes up for editorial weakness: it amplifies what already works.
How to Implement It in 4 Steps
- Create the file in Markdown using the standard structure (H1, blockquote, H2 sections, annotated links)
- Save it as
llms.txtand upload it to the domain root - On WordPress, plugins such as Website LLMs.txt or AIOSEO automate the generation
- Consider creating versions
.mdonly for the most strategic pages, avoiding duplicates that harm traditional SEO
FAQ
Does llms.txt improve Google rankings?
No. Google has confirmed that it does not use it in its algorithms. Its impact concerns visibility in AI systems such as ChatGPT and Perplexity.
Is it mandatory?
No. AI models read websites even without it, but with less precision in interpreting priority content.
Do I need to create .md files for every page?
It is not advisable to do this wholesale: it generates duplicate content and risks penalising organic SEO.
Is llms.txt relevant for agencies and consultants?
Yes. Anyone publishing authoritative content on branding and strategy has a direct interest in being cited correctly by the AI systems that decision makers consult every day.
The Next Step for Your Brand
In 2026, digital visibility is played out on several levels. Bliss works alongside entrepreneurs and marketing managers to build brands that are coherent and recognisable within the AI ecosystem too. Contact Bliss for a consultation and discover how to make your brand citable, authoritative and present wherever your customers look for answers.