llms.txt for SaaS: Spec, Template, Implementation (2026)

llms.txt is a markdown-formatted file placed at the root of your website that gives AI engines a curated, structured map of your most important content. It is the AI-era equivalent of robots.txt and sitemap.xml combined, but built specifically for large language models. As of 2026, publishing one is becoming a baseline expectation for any B2B SaaS that takes AI search visibility seriously. The file takes thirty minutes to create, costs nothing, and signals to AI engines that your site is AI-aware and AI-organized.

What llms.txt actually is and where it came from

llms.txt is a proposed convention introduced in 2024 by Jeremy Howard, the founder of Answer.AI and a well-known figure in machine learning. The convention proposes a single markdown file placed at https://yourdomain.com/llms.txt that contains a brief site description, followed by lists of your most important pages organized into sections, with each page entry showing the title, URL, and a one-line description.

The file is not a W3C standard. It is an emerging convention adopted by AI engines and crawlers as a useful signal that the site is structured for machine consumption. As of 2026, Anthropic’s Claude, Mistral, and a growing number of crawlers actively fetch llms.txt when they discover a domain. OpenAI and Google have not officially confirmed support but their crawlers fetch the file when present, which suggests they at minimum log its existence.

The goal of the file is simple: give AI engines a curated map of your best content rather than making them infer it from your sitemap. A sitemap lists every URL on your site, indexed by date. An llms.txt lists your fifteen to thirty highest-value pages, organized by section, with human-written descriptions that explain what each page covers and why it matters.

Why B2B SaaS should publish llms.txt now

Four reasons make llms.txt a clear yes for any serious SaaS in 2026.

First, AI engines are starting to read it. The reading is not yet uniform across all engines, but the trend is one direction — more engines reading the file over time, not fewer. Getting ahead of the curve costs nothing.

Second, having an llms.txt is itself a signal of AI-readiness. Anecdotally, sites with well-structured llms.txt files report better AI citation rates over time, though it is difficult to isolate this as a single causal factor.

Third, the file is trivial to implement. A 1-to-2-kilobyte markdown file. No code. No risk of breaking anything. The implementation time is less than the time you spend deliberating about whether to do it.

Fourth, there is zero downside. If AI engines ignore your file, nothing breaks. If they read it, you have shaped the citation pool by telling them which pages to prioritize. The risk-reward asymmetry is heavily in favor of publishing.

The llms.txt specification

The proposed spec is a markdown file with three required components: a header line with your brand name as an H1, a blockquote with a one-to-two-paragraph description of the site, and one or more H2 sections each containing a list of links with descriptions.

The general structure looks like this:

# Site Name

> A concise description of what the site is, who it is for, and what makes it credible. This appears prominently to AI engines reading the file.

## Section Name (e.g., Core services, Resources, Tools)
- [Page Title](https://yourdomain.com/page/): One-line description of what the page covers and why it matters.
- [Another Page](https://yourdomain.com/another/): Description.

## Optional Section
- [Page](url): Description.

Template specifically for B2B SaaS

For a typical B2B SaaS, the file should be organized into sections that reflect how a buyer would explore your site: what the product does, who it serves, how it compares to alternatives, supporting evidence, and educational content. Here is a template you can adapt:

# [Your SaaS Brand]

> [One paragraph: what the product is, who buys it, what makes it credible. Mention category, customer profile, traction proof (customer count, funding stage, notable customers).]

## Core product
- [Product Overview](url): What the product does, key capabilities, who it is for.
- [Pricing](url): Pricing tiers and what is included at each level.
- [Use Cases](url): Top use cases by buyer persona or job function.

## For evaluators
- [Comparison vs Competitor A](url): Honest side-by-side comparison with feature matrix.
- [Comparison vs Competitor B](url): Same for second main competitor.
- [Case Studies](url): Customer outcomes with named brands and specific metrics.
- [Security & Compliance](url): SOC 2 status, GDPR, technical security architecture.

## Educational
- [Definitive Guide to X](url): Cornerstone educational content for your category.
- [How to Approach Common Buyer Question](url): Tactical guide for a top-of-funnel question.

## Company
- [About](url): Team, mission, investors, history.
- [Blog](url): Latest thinking and analysis.

How to implement llms.txt on WordPress

Three implementation paths, ranked from easiest to most controlled.

Option one: Yoast SEO Premium. Recent versions of Yoast Premium auto-generate llms.txt based on your cornerstone content settings. Enable the toggle in Yoast settings under Tools. This works for most sites but gives you less control over which pages appear in the file.

Option two: Static file upload. Write a markdown file in any text editor, then upload it to the root of your web server at /var/www/html/llms.txt via SFTP or your hosting control panel. Make sure your web server is configured to serve .txt files at the root — most are by default. This is what most teams should do because it gives full control over the file contents.

Option three: WordPress filter or rewrite rule. Add a rewrite rule that dynamically generates the file based on your published pages and their custom fields. This requires custom PHP work but updates automatically when you publish new content.

Tools and resources that help

For drafting the file: llmstxt.org hosts the spec and example files from other sites. Llms.txt Hub on GitHub catalogs published files you can learn from. Markdown Live Preview or any markdown editor (Typora, Obsidian, VS Code with preview) lets you preview formatting. For verification that AI engines are reading it: server log analysis for user agents like ClaudeBot, PerplexityBot, GPTBot fetching /llms.txt confirms reads. For ongoing maintenance: set a calendar reminder quarterly to review and update the file as your site evolves.

Common llms.txt mistakes

Listing every page on your site. llms.txt is curation, not a sitemap. Pick fifteen to thirty highest-value pages, not every URL you have ever published. Generic, useless descriptions like “Our service page” — each description should tell the AI what the page actually covers in one specific sentence. Stale URLs that point to pages that no longer exist look worse than no file at all. Quarterly refresh is realistic. No header description, or a generic one — the opening paragraph is your entity-positioning statement, invest in it. Confusing llms.txt with robots.txt — robots.txt restricts crawlers from accessing parts of your site, llms.txt invites and curates which parts they should prioritize. They serve opposite purposes.

What llms.txt is not

It is not a guarantee of AI citation. Publishing the file does not directly cause AI engines to cite you. It is one signal among many that influence citation likelihood. It is not a replacement for content quality or entity authority — if your underlying content is weak or your brand is unknown, llms.txt cannot fix that. It is not a blocker for AI engines that do not read it. They will still crawl your site normally through your sitemap and homepage. llms.txt is a complement that compounds with the other six pillars of AI search authority, not a standalone solution.

Frequently asked questions

How long does an llms.txt file need to be?

Typically 1 to 3 kilobytes — roughly 30 to 80 lines of markdown. If you find yourself approaching 10 kilobytes you are listing too much; cut to your most important pages.

Should I list my pricing page in llms.txt?

Yes. Pricing is one of the highest-intent pages buyers reach. Having it explicitly listed with a clear description helps AI engines surface it when buyers ask pricing questions.

Will llms.txt affect my Google rankings?

No direct effect. Google’s traditional ranking algorithm does not weight llms.txt as a signal. It is purely an AI-engine signal at this stage.

Do I need to submit my llms.txt anywhere?

No submission process exists. AI engine crawlers discover it by fetching the standard path. Make sure the file is publicly accessible at /llms.txt from your root domain.

llms.txt for SaaS: The 2026 Spec, Template, and Implementation Guide