Startup SEO·6 min read·2026-03-12

How to write an llms.txt file (and why every startup needs one)

SEO Strategist
In brief: An llms.txt file is a markdown document hosted at your website's root that tells AI tools what your business does, how your product works, and where to find key information. It takes 20 minutes to create and is one of the simplest ways to improve your visibility in AI-generated search results from ChatGPT, Perplexity, and Google AI Overviews. Last updated: March 2026

Your website has a robots.txt. It probably doesn't have an llms.txt.

Most founders know about robots.txt. It tells search engine crawlers what to index. Sitemaps tell Google where your pages are. But neither of these files is designed for the systems that are rapidly becoming the primary way people find products: large language models.

When someone asks ChatGPT "What's a good AI tool for startup branding?" the model doesn't crawl your site in real time. It relies on training data, retrieval-augmented generation, and whatever structured information it can access quickly. If your site is a mess of JavaScript, marketing fluff, and nested navigation, the model has to guess what you do. Often, it guesses wrong. Or worse, it ignores you entirely.

That's the problem an llms.txt file solves.

What is an llms.txt file?

An llms.txt file is a plain markdown document, proposed by AI researcher Jeremy Howard, that provides LLMs with a concise, structured summary of your website. Think of it as a cover letter for AI: here's who we are, here's what we do, here's where to find the detail.

The file sits at your domain root (yoursite.com/llms.txt) alongside robots.txt and sitemap.xml. Unlike those files, llms.txt is written in markdown because that's the format language models parse most easily. The specification is maintained at llmstxt.org and has been adopted by companies including Stripe, Cloudflare, and dozens of documentation platforms.

According to research from GEO firm Brandlight, the overlap between top Google results and AI-cited sources has dropped below 20%. Your site might rank well on Google and still be invisible to AI search. An llms.txt file won't fix everything, but it gives AI systems a clean signal of what you are and what matters on your site.

The format

Every llms.txt file follows a simple structure:

# Your Company Name


> A one-sentence description of what you do.



More detail about your product, audience, and market position.



Section Name




Another Section
The file starts with an H1 (your company name), an optional blockquote summary, then sections organised by topic. Each section lists key pages with URLs and short descriptions. That's it.

No HTML. No JSON. No schema markup. Just clean markdown that a language model can read in a single pass.

Step-by-step: writing yours

Step 1: Define your core identity in 2-3 sentences

Write a description of your product as if you were explaining it to a smart colleague who has never heard of you. Avoid marketing language. Be factual and specific.

Bad: "We're a next-generation platform empowering founders to achieve their full potential."

Good: "Flint is an AI workspace where solo founders get a team of six AI specialists covering brand strategy, pitch decks, SEO, social media, copywriting, and launch planning. Each specialist replaces a hire."

Step 2: Map your most important pages

Not every page belongs in llms.txt. Include the pages that answer the questions AI systems are most likely to encounter about you. For a typical startup, that's your homepage, product pages, pricing, about page, and 5-10 of your best content pieces.

Step 3: Organise into sections

Group pages logically. Common section headers include "About," "Products," "Documentation," "Blog," and "Legal." Use whatever structure reflects how a knowledgeable person would think about your site.

Step 4: Write brief descriptions for each link

Each page gets a one-line description. These descriptions matter because they tell the AI why a page is relevant, not just that it exists. Compare:

Weak: - Blog: Our blog

Strong: - SEO for Startups: Complete guide to building search visibility from zero, covering technical SEO, content clusters, and GEO

Step 5: Host it at your root

Upload the file to your domain root so it's accessible at yoursite.com/llms.txt. If you're using Next.js, drop the file into your public/ directory. For WordPress, upload via your host's file manager or use a plugin. For platforms like HubSpot or Squarespace that don't natively support root file hosting, you may need a redirect from /llms.txt to a hosted file URL.

Verify it works by visiting the URL directly. You should see raw markdown.

A real example

Here's a trimmed version of the llms.txt file we built for Flint:

# Flint


> Your AI startup team. One workspace for solo founders who move fast.



Flint is an AI-native workspace where solo founders get a full startup team. Six AI specialists cover brand, pitch decks, SEO, social, copy, and launch. Built by RightsTech Ltd (UK).



Products


  AI-powered brand strategy from positioning to visual identity
  Investor-ready pitch decks with narrative structure


Key Concepts



  • Artifact Drive: shared project folder across all team members
    • 

  • Business Brief: shared context document all specialists read from
    • 

  • Onboarding: each specialist learns your business before executing
Twenty lines. Took 15 minutes to write. Now every AI system that accesses runflint.ai/llms.txt gets a clean, accurate summary of what Flint does.

What llms.txt won't do

It's worth being honest about the limits. Google's John Mueller compared llms.txt to the keywords meta tag, implying it may have limited direct impact on rankings. Not all AI crawlers check for llms.txt yet. And having the file doesn't guarantee citation.

But the cost of creating one is close to zero, and the potential upside grows every quarter as AI search traffic increases. ChatGPT now reaches over 800 million weekly users, according to OpenAI. Google AI Overviews appear in at least 16% of searches, according to Search Engine Land. If even a fraction of that traffic reaches your llms.txt, it's time well spent.

Pair it with the rest of your GEO stack

An llms.txt file is one piece of a broader generative engine optimisation strategy. For maximum AI visibility, combine it with structured FAQ sections on every page, citation-worthy statements containing specific facts and numbers, in-text source attribution that AI systems can reference, and clean HTML that AI crawlers can parse without JavaScript rendering.

If you're building search visibility from scratch, Flint's SEO Protocol builds a complete keyword strategy, content plan, and GEO audit from your domain. It's one of six AI specialists in the workspace. Join the waitlist if you'd rather not do this manually.

Frequently asked questions

What is an llms.txt file?

An llms.txt file is a markdown document hosted at your website's root directory that provides large language models with a structured summary of your site. Proposed by AI researcher Jeremy Howard and documented at llmstxt.org, it helps AI tools like ChatGPT and Perplexity understand your business without parsing complex HTML.

Is llms.txt the same as robots.txt?

No. robots.txt controls which pages search crawlers can access. llms.txt provides context about your site's content and structure to help AI models understand and reference your business accurately. You need both files.

Does Google use llms.txt for rankings?

Google's John Mueller has compared llms.txt to the keywords meta tag, suggesting limited direct impact on traditional rankings. However, AI-powered search features like Google AI Overviews draw from structured, accessible content, and llms.txt contributes to that accessibility.

How long should an llms.txt file be?

Most effective llms.txt files are 30-100 lines. Include your core identity, key product pages, and top content. Don't list every page on your site. Focus on the 10-20 URLs that best represent what you do.

Do AI tools actually read llms.txt?

Adoption is growing. SEMrush now checks for llms.txt presence in site audits. Multiple AI crawlers request the file when indexing sites. While not universal yet, the trend is clear: as AI search grows, structured AI-readable files become more valuable.

How often should I update my llms.txt?

Update it when you launch major features, add new product lines, or restructure your site. For most startups, quarterly reviews are sufficient. The file should reflect your current product and content, not historical pages.

Sources