llms.txt examples: what 57 B2B SaaS companies put in theirs

llms.txt is a Markdown file at the root of a website, such as example.com/llms.txt, that gives AI tools a short reading list of the site's most important pages. It was proposed in 2024 and is now one of the most talked-about files in SEO. So I checked what 57 well-known B2B SaaS companies actually put in theirs.

On 21 October 2026 I fetched the llms.txt, llms-full.txt and robots.txt files of companies such as HubSpot, Stripe, Atlassian, Notion, Zendesk and Vercel. Below are the numbers, five real patterns with examples, what the file does and doesn't do, and a template for a SaaS site. The full results table is at the end.

I work at Fourward, which publishes this study and sells SEO, AEO and GEO work. No company in the study knew about it. Every file is linked in the results table so you can check it yourself. Files change often, so what you see may differ from what I saw on 21 October 2026.

What llms.txt is

llms.txt was proposed by Jeremy Howard in September 2024. The specification at llmstxt.org describes a Markdown file that provides "LLM-friendly content": a quick guide an AI tool can read instead of crawling a whole site.

The format is simple:

• H1 titleThe name of the site or project. It's the only required part.
• SummaryA short blockquote that says what the site is.
• DetailsOptional paragraphs or lists with more context.
• Link listsSections with H2 headings, each a list of links with a short note.
• OptionalA section titled "Optional" for links an AI tool can skip when it needs less context.

The spec also suggests offering a clean Markdown copy of each important page at the same address plus .md, and some sites add an llms-full.txt file with the full text of their docs.

Does llms.txt help SEO or AI visibility?

For Google, no. Google's guide to generative AI features in Search says you don't need "AI text files, markup, or Markdown to appear in Google Search", that "Google Search itself doesn't use them", and that keeping one for other services "will neither harm nor help" your visibility there.

Google's Chrome team does check for it, but for a different reason. Lighthouse has an llms.txt audit in its agentic browsing category, aimed at AI agents that use websites. It only flags a server error when the file is requested. A missing file is marked not applicable, "as providing the file is optional at the moment".

So llms.txt is mainly for AI agents and assistants that read a site while doing a task, such as coding tools reading developer docs. That's why developer-tool companies have the most detailed files in this study.

How I checked

I picked 60 well-known B2B SaaS companies across CRM, collaboration, developer tools, payments, analytics, support and HR. For each, on 21 October 2026, I requested:

• /llms.txt and /llms-full.txt on the main domain, with and without www.
• /llms.txt on the docs. subdomain, following redirects.
• /robots.txt on the main domain.

I counted a file only if it came back as text, not an HTML page, and I used a normal browser user agent. Three companies are left out: Segment now redirects to Twilio, and Gusto and Contentful blocked my requests. That leaves 57.

Sizes are in kilobytes. "Links" counts Markdown links in the file. "Links to .md" means the file points to Markdown versions of pages.

What 57 B2B SaaS companies put in llms.txt

The sections companies choose say a lot about what they want AI to know. Of the 47 main-site files:

• Blog, resources or guides40 files have a section for them.
• Products, platform or features34 files.
• Docs, API or developers30 files.
• Solutions, use cases or industries27 files.
• Pricing or plans21 files.
• An "Optional" section17 files, as the spec suggests.
• Customer stories15 files.
• Comparisons or alternatives10 files.

Three files break the spec's one hard rule by not starting with an H1 title: Algolia's starts with a plain line of text, and Salesforce's and Twilio's start straight into a list of links.

Five llms.txt patterns, with real examples

Most files fit one of five shapes. Each one suits a different kind of company.

1. The short index

Asana (6 KB, 44 links), Notion (7 KB, 48 links) and Typeform (5 KB, 23 links) open with an H1 and a short summary, then group their key pages: core products, use cases, solutions by team and support. Pipedrive's (9 KB, 43 links) is the best B2B example I found: it has sections for pricing, integrations, who Pipedrive is for and comparisons with other CRMs. This pattern is the closest to what the spec describes, and it's what I'd recommend for most SaaS marketing sites.

2. The docs index

Developer tools point AI straight at Markdown. In Linear's file, 129 of 131 links go to .md pages, such as its start guide and concepts. Cloudflare (82 of 90) and Vercel (20 of 28) do the same. Cloudflare also publishes a separate file on its developer docs site.

3. The marketing map

HubSpot's file (35 KB, 225 links) is a map of the whole company. It explains how the product hubs fit together, then lists each product with a one-line description. Atlassian adds pricing, deployment options, security and integrations. Gong lists solutions by role and industry, integrations and customer case studies. Semrush even has a section on its competitive positioning.

4. The full dump

Some files list nearly every page on the site. Twilio's is 2.2 MB with more than 9,000 links. Salesforce's is almost 1 MB, and Zendesk's, Datadog's and Mailchimp's are each over 500 KB. A file that size is a sitemap in Markdown. It defeats the point of a short reading list, and an AI tool with limited context may only read the top of it.

5. Instructions for agents

A few files tell AI agents what to do, not just what to read. Loom's file, under 1 KB with no Markdown links, says Loom is part of Atlassian and asks AI agents to connect to Atlassian's MCP server "rather than crawling this site". Vercel's has sections on when to use Vercel and how agents should use it.

llms.txt and robots.txt are separate

An llms.txt file doesn't let AI in. robots.txt decides which crawlers may read your site, and some companies publish llms.txt while blocking AI bots.

• LoomPublishes llms.txt but blocks GPTBot, OpenAI's training crawler, in robots.txt.
• FigmaHas no llms.txt, blocks the training crawlers, and gives AI search crawlers an allowlist of specific pages in its robots.txt.
• Five moreAmplitude, Calendly, GitHub, Notion and Postman publish llms.txt and fully block at least one AI-related bot, such as Bytespider, CCBot or Meta's crawler.

None of the 57 blocks the search crawlers behind ChatGPT, Claude or Perplexity. That's the setting that matters for AI answers, and it's covered step by step in our GEO guide for B2B SaaS.

An llms.txt template for a B2B SaaS site

This is the short-index pattern, adapted for a SaaS marketing site. Replace the names and links, keep each note to one line, and stay under about 50 links.

# Acme

> Acme is a customer onboarding platform for B2B SaaS teams. It helps
> customer success teams build in-app checklists, track activation and
> hand off accounts from sales. Plans start at $99 a month.

Acme is used by SaaS companies with 20 to 2,000 employees. It integrates
with Salesforce, HubSpot, Segment and Slack.

## Product
- [Platform overview](https://acme.com/product): What Acme does and how it works
- [Onboarding checklists](https://acme.com/product/checklists): In-app checklists and tours
- [Activation analytics](https://acme.com/product/analytics): Track which accounts activate

## Pricing
- [Pricing](https://acme.com/pricing): Plans, limits and what each includes

## Compare
- [Acme vs Competitor](https://acme.com/compare/competitor): Side-by-side comparison
- [Competitor alternatives](https://acme.com/alternatives/competitor): Why teams switch

## Integrations
- [All integrations](https://acme.com/integrations): Salesforce, HubSpot, Segment, Slack

## Company
- [About](https://acme.com/about): Team, founding year and investors
- [Customers](https://acme.com/customers): Case studies with results
- [Security](https://acme.com/security): SOC 2 report and data policies

## Optional
- [Blog](https://acme.com/blog): Guides on onboarding and activation
- [Changelog](https://acme.com/changelog): Product updates

Five rules I'd follow

1. Lead with plain facts. Say what the product is, who it's for and what it costs in the summary, the same way your homepage does.
2. Link canonical pages only. No tracking links, no redirects, no pages blocked in robots.txt.
3. Include pricing and comparisons. They answer the questions buyers ask AI most, and fewer than half the files in this study include them.
4. Serve it as text. Return text/plain or text/markdown with a 200 status, not an HTML page.
5. Keep it current. Update it when you change pricing or launch a product. An out-of-date file is worse than none.

Should your SaaS company add one?

Our free AI crawler checker tests whether AI bots can reach your site and whether your llms.txt loads. The AI visibility check looks at the wider picture, from structured data to comparison pages.

Full results: 57 B2B SaaS companies

Checked 21 October 2026. "Yes" links to the file. Scroll the table sideways on a phone.

FAQs

Sources

1. The /llms.txt file, llmstxt.org, proposed by Jeremy Howard, September 2024.
2. Google's guide to optimizing for generative AI features on Google Search, Google Search Central, updated July 2026.
3. llms.txt audit, Lighthouse, Chrome for Developers, May 2026.
4. The llms.txt, llms-full.txt and robots.txt files of the 57 companies in the results table, checked 21 October 2026.