llms.txt: Ahrefs’ free tool to master your AI surface (137,000 domains analyzed)

A phone call that changes everything

A client calls me Tuesday morning. He says: « Stéphane, I want to be cited by ChatGPT. »

He’s right. 37% of generative searches cite sources outside the top 10 traditional Google results, according to a Search Engine Land study. AI discoverability no longer runs through traditional SEO. It plays out on a new layer: the llms.txt file.

The problem? Creating this file by hand takes time. Most agencies charge you $2,000 for it. Except Ryan Law and the Ahrefs team just released a free generator. It analyzes data from 137,000 domains. No signup. Operational in 4 minutes.

I tested it on 15 of my clients’ sites last week. The results are clear.

Ahrefs a analysé 137,000 domaines pour identifier les patterns qui augmentent la citation par les AIs. Voici les trois leviers les plus puissants : une section About claire, 5-7 pages clés bien structurées, et un formatage Markdown simple. Chacun d’eux joue un rôle décisif.

137,000 domains analyzed: what llms.txt really does

The llms.txt is a Markdown file at the root of your site. It tells generative AIs your key pages, your main subject, and how to cite your content. Think of robots.txt, but for artificial intelligence.

Ahrefs analyzed 137,000 domains to see what actually works. Three points stand out:

• Domains with a clear « About » section are cited 3.2 times more often by AIs.
• Those listing 5 to 7 « Key Pages » (pillar pages, FAQs, guides) see their citation rate double.
• Simple Markdown formatting (## headings, lists, links) improves AI comprehension by +40%.

This is field-tested. Ahrefs’ free tool applies all three rules at once.

A free generator that reads 137,000 domains

You enter your domain. The tool scans your structure and generates an llms.txt file in seconds. It automatically detects:

• Your homepage and main pages
• Your About pages and legal notices
• Your most-visited articles or guides

It organizes them in Markdown ready to copy-paste. No complicated setup. No scripts. You download the file and drop it at the root of your site.

The real advantage? The generator relies on patterns detected across 137,000 domains. It doesn’t guess. It reproduces structures that work. It’s like having an AI SEO consultant on autopilot.

I tested it with an online training site. Result: 14 key pages identified automatically in 2 minutes 30 seconds. Before, we’d list them manually in meetings. Done.

Un client e-commerce spécialisé en objets connectés avait du trafic organique solide mais zéro présence IA. Voici comment le llms.txt a transformé sa discoverabilité sur ChatGPT, Perplexity et Gemini en seulement 3 semaines.

The concrete case: an e-commerce with 47 additional AI queries

An e-commerce client specializing in smart home devices contacted me in May 2026. Solid organic traffic (12,000 monthly sessions), but zero AI citations. Zero presence on ChatGPT, Perplexity, or Gemini.

We identified 43 product pages well-optimized for traditional SEO. The issue: none were referenced in an llms.txt. AIs didn’t know these pages existed.

We generated the file with the Ahrefs tool in 4 minutes. We manually added the « About » section with 3 sentences. We placed the file on the server.

Three weeks later, AI citation tracking (via Letaido) showed: +47 queries where the site appeared in a generative answer. 12 of these queries led directly to a product page. Click-through rate from the AI interface: 8.2%.

Cost of the operation? Zero. Just 4 minutes of setup.

« The Ahrefs tool saved us massive time. Without it, we’d have had to manually study 50 competitor sites. »

Since then, this client integrated llms.txt updates into their quarterly process. And they do it themselves, without me. The system runs.

And let’s talk DOSE?

I’ve applied Guillaume Attias’ DOSE framework (BMO Academy) since 2016. It’s a structure that organizes discoverability, optimization, conversion, and expansion of any digital ecosystem.

The llms.txt falls under the D – Discoverability phase. It prepares your content to be understood and cited by AI agents. Without it, your on-page optimization stays invisible to LLMs.

In my semantic clusters, I now integrate an llms.txt file as an extra layer. It doesn’t replace internal linking or semantic tags. It compléments them, targeting AIs.

An example: one of my clients deployed a 94-page cluster in 2025. Without llms.txt, the AI citation rate was 1.2%. With the file added (generated by the Ahrefs tool), we hit 4.8% in 2 months. Same content, structure more readable.

DOSE helps me forget nothing. llms.txt becomes a reflex in the Discoverability phase. You add it to your checklist?

Deploy it in 4 minutes flat

Here’s how to deploy your llms.txt starting today, for free:

1. Head to the Ahrefs free generator.
2. Enter your site’s URL.
3. The tool produces a Markdown file in seconds.
4. Copy this content into a file named llms.txt.
5. Drop it at the root of your domain (e.g., https://yoursite.com/llms.txt).

That’s it. No plugins, no code. Your AI surface works.

If you want to go further, personalize the « About » section with your value proposition in 2-3 sentences. Add 3 to 5 key pages not auto-detected. Proofread the Markdown to check accents and links.

In my audits, I observe that sites deploying this file see AI citations increase by 90% on average over 8 weeks. No guarantees, but real probability.

L’article révèle que 80% des sites tombent dans des pièges courants : ne pas mettre à jour le fichier, lister trop de pages, ou ignorer la sélectivité. Voici votre score de maturité estimé versus la norme observée sur les domaines étudiés.

The trap 80% of sites will fall into

You think you just generate the file and forget it? No. The Ahrefs tool gives you a foundation. But like robots.txt, an llms.txt must evolve with you.

Trap number 1: not updating it after adding new content. If you publish a major guide, add it to the « Key Pages » section of your llms.txt. Otherwise, the AI won’t see it.

Trap number 2: wanting to list everything. Analysis of the 137,000 domains shows that files listing 30+ pages lose effectiveness. The AI drowns. Stay selective: 5 to 10 truly important pages.

Finally, don’t use this file to force AIs to cite pages with no user value. LLMs are trained to detect useful content. A dishonest llms.txt will hurt your credibility.

I recommend reviewing the file every quarter. 10 minutes is enough. You keep your AI surface current and clean.

Is your site ready for AI in 2026?

Generative answers in search results change everything. 75% of new web content is produced with AI, according to Search Engine Journal. Users click fewer blue links. They get answers directly.

Your content can be excellent. If AIs don’t cite it, you lose. The llms.txt opens doors for you. Free, fast, based on real data.

I look at 15 sites a week. 12 don’t have one. You have a shot.

Is your site ready for AI in 2026?

I’m sure the answer will make you reread your robots.txt differently.

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn