What Is llms.txt and Do You Need One?

If you follow GEO or AI search, you've probably seen llms.txt recommended as a must-do tactic. Some of the marketing around it borders on breathless. This guide takes the opposite approach: real data, honest assessment, practical instructions. We use llms.txt on this very site — and we've built a free generator tool — so we have a genuine interest in the file being useful. That makes us more motivated, not less, to tell you exactly where the evidence stands.

What is llms.txt?

llms.txt is a plain Markdown file placed at the root of your website — at yoursite.com/llms.txt — that gives AI systems a curated, structured summary of your site. It was proposed in September 2024 by Jeremy Howard (co-founder of Answer.AI and fast.ai).

The core idea: modern web pages are noisy. Navigation menus, cookie banners, JavaScript, ads, popups — all of it makes it harder for an AI model to quickly understand what your site is about and which pages matter most. An llms.txt file cuts through that noise by providing a clean, machine-readable summary in one place.

Think of it as a guided entry point. It doesn't control access (that's what robots.txt does) — it tells AI systems "here's who we are, here's what matters, and here's where to look."

The format: what goes in the file

The format is deliberately simple. A valid llms.txt contains:

• A single H1 heading with your site or brand name — the only strictly required element.
• A blockquote summary (optional but strongly recommended) — one or two sentences describing what you do and who you're for. This is arguably the most important line: it's what AI models read first to understand your identity.
• Context paragraphs (optional) — a sentence or two of extra detail.
• H2 sections with link lists — each section groups your key pages with short descriptions, formatted as Markdown links.

Here's a real-world example — this is roughly what our own llms.txt looks like:

# Visibly

> GEO platform that tracks brand visibility across AI search engines
> — ChatGPT, Gemini, Perplexity and Claude — and delivers
> prioritised recommendations to get recommended more.

Visibly is built for in-house marketing teams at mid-size brands
watching organic traffic decline as AI intercepts customer journeys.

## Key pages

- [Home](https://getvisibly.co.uk/): What Visibly does, pricing, early access
- [What Is GEO?](https://getvisibly.co.uk/blog/what-is-geo): Complete guide to Generative Engine Optimization
- [Is SEO Dead?](https://getvisibly.co.uk/blog/is-seo-dead): Why organic traffic is declining

The file must be named exactly llms.txt (lowercase), saved as UTF-8, and placed at your domain root so it's accessible at yoursite.com/llms.txt.

llms.txt vs robots.txt vs sitemap.xml

These three files are related but serve different purposes, and understanding the distinction matters:

The key distinction: robots.txt lets crawlers in. sitemap.xml tells them what exists. llms.txt tells AI what matters. They're complementary, not competing — you need all three for a well-structured site.

Real adoption data

Here's where most articles get vague. The data, as of mid-2026, is actually quite clear:

Overall adoption is still niche. An SE Ranking study of 300,000 domains found 10.13% had an llms.txt file.¹ A separate crawl of the Tranco Top 10,000 found roughly 5.6%, growing from about 1% a year earlier.² An Ahrefs study of 137,000 domains found 28% — but noted their sample skews heavily toward tech-savvy, AI-aware sites, making the true web-wide rate much lower.³

Shopify is a major outlier. In late April 2026, Shopify silently pushed llms.txt to every store on the platform by default — no merchant opt-in, no announcement. Their 78% adoption rate is a platform decision, not individual enthusiasm.²

Early adopters are developer-facing. The companies that adopted first — Anthropic, Stripe, Cloudflare, Vercel, Cursor — are developer-documentation sites, not consumer brands. This pattern matters, because it reveals where the file is genuinely being consumed (coding tools), versus where it's aspirational (marketing sites).

Does it actually improve AI visibility?

This is the question that matters, and I'm going to give you the honest answer rather than the marketing one: no, not measurably, as of mid-2026.

The SE Ranking study found no correlation between having an llms.txt and being cited more frequently by AI engines. Removing the variable from their machine-learning model actually improved its accuracy — meaning it was adding noise, not signal.¹

An ALLMO analysis of 94,000+ cited URLs found the file in less than 1% of domains that AI engines actually cited — far below the 10% adoption rate, suggesting no uplift.⁴

The Ahrefs study was perhaps the most striking: of the roughly 38,000 domains in their dataset with a valid llms.txt, 97% received zero requests for the file in May 2026. No bots, no humans, nothing.³

Google's own position is ambiguous: its new guide on optimising for generative AI features explicitly listed llms.txt under "mythbusting" as not needed, while its Chrome team simultaneously added an llms.txt check to Lighthouse experimental audits.³

Should you add one?

Yes — but for the right reasons, with realistic expectations. Here's the honest case:

The cost is near zero. Creating an llms.txt takes minutes (especially with our free generator). Deploying it is one file upload. The ongoing maintenance is minimal — update it when you add or remove major pages.

There's no downside risk. No study has found that having an llms.txt harms visibility or causes any negative outcome. The worst case is it sits there doing nothing.

The forward-compatibility argument is strong. Structured, machine-readable signals have a track record on the web. Schema markup took years to matter and is now expected. robots.txt was proposed in 1994 and every major crawler respects it today. If llms.txt follows a similar arc — and several signs suggest it might — being already correct when that happens is cheap insurance.

The exercise itself has value. Curating a one-paragraph summary and a short list of your most important pages forces you to answer "what would I want an AI to say about us?" Most marketing teams haven't done that exercise, and the clarity it produces is useful regardless of whether the file is consumed.

What you should not expect: a measurable lift in AI citations or search traffic this quarter. If someone tells you llms.txt is a ranking factor, they're selling you something the data doesn't support.

How to create and deploy yours

Two paths, depending on how much time you want to spend:

The fast way (under 2 minutes)

Use our free AI llms.txt generator. Paste your website URL, and it reads your site and drafts a valid file for you. Edit anything that needs adjusting, download, and deploy. No signup needed to generate.

The manual way (15–30 minutes)

Create a new file called llms.txt in any text editor. Follow this structure:

# Your Brand Name

> One or two sentences describing what you do and who you're for.
> Write this as you'd want an AI to summarise you — factual, specific, no hype.

Optional extra context paragraph.

## Key pages

- [Page title](https://yoursite.com/page): Short description of what this page covers
- [Another page](https://yoursite.com/other): What this one covers

Not sure what “good” looks like when it’s finished? We’ve published four complete llms.txt examples — a service business, a SaaS, a docs site and an online shop — with the mistakes to avoid.

Save it, then upload it to the root of your website — the same folder as your homepage — so it lives at yoursite.com/llms.txt. Visit that URL in your browser to confirm it loads as plain text. That's the entire deployment.