How many SaaS products publish llms.txt?

Nearly two-thirds of the SaaS products we checked do not publish an llms.txt file.

On August 24, 2026, we checked 17,218 products in our catalog. Of those, 6,876, or 39.9%, had an accessible llms.txt file.

Among the 100 highest-ranked products in our catalog, adoption was substantially higher:64%published one. That is 24.1 percentage points above the catalog-wide rate, making the top 100 roughly 1.6 times as likely to have adopted the format.

What llms.txt actually does

An llms.txt file is a structured Markdown guide to a website. It can summarize the product, explain how its content is organized, and point agents toward important resources such as documentation, API references, tutorials, pricing pages, or changelogs.

The current proposal describes it as a curated entry point that agents can consult before following links to more detailed, preferably Markdown-formatted pages. It is designed to complement tools such as sitemap.xml and robots.txt, not replace them.

That distinction is important. The 39.9% result does not mean the remaining 60.1% of products are invisible to ChatGPT, Perplexity, or other answer engines. A site can still be crawled, indexed, retrieved, and cited without publishing llms.txt. Conversely, publishing the file does not guarantee inclusion or a higher ranking.

OpenAI’s official guidance for appearing in ChatGPT search focuses on allowing OAI-SearchBot, while Perplexity recommends allowing PerplexityBot. Both are managed through robots.txt; neither company identifies llms.txt as a search-ranking requirement.

Our result is therefore best understood as a measure of AI-oriented publishing readiness, rather than a direct measure of answer-engine visibility.

Adoption is higher among prominent products

The gap between our overall catalog and its top 100 products is the most interesting part of the data.

Across the full sample, adoption was 39.9%. Among the top 100, it reached 64%. Products in that group publishing llms.txt include Vercel, Replit, OpenRouter, Semrush, Frappe Builder, n8n, Kiro, and WooCommerce.

The list skews toward developer tools, infrastructure platforms, and products with extensive technical documentation. That is consistent with how the format is commonly used: the llms.txt proposal notes that adoption has been particularly useful for software documentation, where agents need to locate precise API references and tutorials.

The correlation should not be mistaken for causation. Publishing llms.txt does not necessarily make a product rank higher in our catalog. Higher-ranked products may simply have more mature documentation, larger technical teams, or documentation platforms that generate the file automatically.

Even with that caveat, the difference is meaningful. More prominent products are considerably more likely to give agents a curated path through their content.

Full-content files are less common

We also detected an llms-full.txt file on 3,506 products, or 20.4%of the sample.

Where llms.txt generally acts as a concise index, llms-full.txt is commonly used as a larger Markdown export containing substantially more of a documentation site’s content. Some documentation platforms generate both files automatically.

The current v2 llms.txt proposal emphasizes a compact guide followed by links to relevant Markdown pages that agents can retrieve when needed. That makes llms-full.txt better treated as a companion implementation pattern than as a separate requirement for compliance.

Presence is only the first test

Our scan measured whether a file existed. It did not evaluate whether the file was accurate, complete, current, or useful.

A strong llms.txt file should:

• Explain clearly what the product does.
• Link directly to the most authoritative pages.
• Give each link a useful, specific description.
• Prioritize documentation, API references, pricing, security, and changelog content where relevant.
• Point to clean Markdown versions of pages when available.
• Be updated when the site or documentation changes.

A file filled with stale links or generic descriptions may technically count as adoption while doing little to help an agent answer questions correctly.

The next useful benchmark is therefore not simply how many products publish llms.txt, but how many publish a good one: valid links, clear descriptions, useful page coverage, current content, and formats that agents can retrieve reliably.

Methodology

We recorded this snapshot on August 24, 2026.

• Catalog entries:17,261
• Products publishing llms.txt:6,876, or 39.9%
• Products publishing llms-full.txt:3,506, or 20.4%
• Top-100 products publishing llms.txt:64, or 64%

The 43 products that could not be checked represented 0.25% of our catalog and would not materially change the overall result.

The scan measures files discoverable at the locations we checked. The current specification permits llms.txt files at subpaths such as/docs/llms.txt, not only at the root of a domain, so the published methodology should state whether documentation subdomains and path-specific files were included.

The results measure file presence, not file quality, crawler usage, search inclusion, ranking, referral traffic, or citation frequency.