What Is llms.txt, and Should You Add One in 2026? By BalochDev · Last updated September 9, 2026 llms.txt is a plain markdown file placed at yourdomain.com/llms.txt that describes your site and lists your most important pages in a format designed for AI agents to read. It was proposed by Jeremy Howard on Answer.AI, September 3, 2024, as a convention to help LLMs understand a site's purpose and content hierarchy without wading through noisy HTML. As of 2026, adoption sits at roughly 10% of domains, and the largest available studies find no statistically significant citation-rate difference between sites with and without it (SE Ranking, 300,000-domain analysis). It does not control crawler access — robots.txt does that — and in June 2026, Google went further than "hasn't confirmed": it stated outright, in its own generative-AI documentation, that llms.txt has no effect on Search rankings or AI Overviews. The honest assessment: llms.txt is a low-cost signal worth adding if your site is complex or context-heavy, but it is not a GEO priority. Fix rendering, crawlers, and schema first. What llms.txt is The convention is simple: a markdown file at your root that describes what your site is, who runs it, and which pages are most important for an AI to understand it. A minimal example: markdown # BalochDev > An AI-first software development studio building products for GCC and international markets. ## Key pages - [About](https://balochdev.com/about): Who we are and what we build - [Blog](https://balochdev.com/blog): Technical guides on AI search, GEO, and software engineering - [Services](https://balochdev.com/services): What we offer - [Contact](https://balochdev.com/contact): Get in touch ## AI Visibility resources - [How to Get Your Business Found by AI Search](https://balochdev.com/blog/how-to-get-your-business-found-by-ai-search) - [5-Minute AI Reachability Check](https://balochdev.com/blog/is-your-website-invisible-to-ai) A fuller version (llms-full.txt) can include complete article content — useful for AI agents that might summarize your site rather than crawl individual pages. What llms.txt doesn't do It does not control crawler access — that's robots.txt. It does not replace schema markup. It does not guarantee any AI engine reads it. It is a convention, not a standard with defined behavior from any major AI company. The evidence against a citation benefit has gotten stronger, not weaker, since the format launched. SE Ranking's analysis of 300,000 domains found 10.13% adoption with no statistically significant citation-rate correlation — and removing the llms.txt variable from their predictive model actually improved its accuracy, meaning the file added noise rather than signal. A separate Ahrefs analysis of 137,000 domains found that 97% of the roughly 38,000 llms.txt files they located received zero requests at all; among the small number that did get traffic, AI bots accounted for only about 1% of requests — most of it came from SEO audit tools and other non-AI bots simply checking whether the file existed. And as of June 15, 2026, Google made its position official: its generative-AI documentation now states plainly that llms.txt has no effect, positive or negative, on Search rankings or AI Overviews. Google's own John Mueller has compared it to the keywords meta tag — a well-known relic from an earlier era of SEO that search engines stopped using as a ranking input decades ago. When it does add value For complex sites with many content types (product, blog, docs, tools all mixed together), llms.txt gives an AI agent navigating your site a useful map — the equivalent of a human-readable sitemap.xml for content purpose rather than URLs. For AI agents performing tasks (agentic browsing, AI coding assistants pulling in documentation), a clear llms.txt can help them identify the right starting point — this is closer to the format's original intent, which was aimed at developer tools and API documentation, not general GEO. Think of it as a clean welcome mat: low cost, minor upside, not a strategic priority. The verdict Add it if it costs you an hour — it signals intentionality, it's genuinely useful for the developer-tool and agentic-browsing use case it was actually designed for, and its value carries some future optionality if a major provider ever does commit to reading it. But don't add it instead of fixing rendering, crawler access, schema, or content structure. Those are the gates that control whether AI can see and cite you at all. llms.txt is decoration on top of a foundation that has to be solid first — and as of mid-2026, the evidence that it moves citation on its own is not just thin, it's actively been tested and found absent. Frequently asked questions Does llms.txt help with Google AI Overviews? No. Google's own June 2026 documentation states it has no effect on Search rankings or AI Overviews. Is llms.txt worth it for a documentation or developer-tool site? More so than for a general marketing site. This is the use case Jeremy Howard originally designed it for — helping coding agents and API-consuming tools navigate documentation efficiently. Anthropic, Stripe, Cloudflare, Vercel, and Supabase all publish one, and that adopter list skews heavily toward developer infrastructure. Will llms.txt matter more in the future? Possibly, if agentic AI browsing grows and a major provider formally commits to reading it. No major provider has done so as of 2026. It carries option value at near-zero cost, which is the strongest argument for adding it — not a current, measurable citation benefit. Sources & further reading Jeremy Howard, /llms.txt — a proposal to provide information to help LLMs use websites — Answer.AI, September 3, 2024 Does llms.txt Actually Work? An Honest Read of the 2026 Evidence — Tobira, May 30, 2026, citing SE Ranking's 300,000-domain analysis 97% Of llms.txt Files Got No Requests, Ahrefs Data Shows — Search Engine Journal, June 16, 2026, reporting Ahrefs' 137,000-domain analysis

AI · Blog

What Is llms.txt, and Should You Add One in 2026?

BalochDev 5 min read

llms.txt is a plain markdown file placed at yourdomain.com/llms.txt that describes your site and lists your most important pages in a format designed for AI agents to read. It was proposed by Jeremy Howard on Answer.AI, September 3, 2024, as a convention to help LLMs understand a site's purpose and content hierarchy without wading through noisy HTML. As of 2026, adoption sits at roughly 10% of domains, and the largest available studies find no statistically significant citation-rate difference between sites with and without it (SE Ranking, 300,000-domain analysis). It does not control crawler access — robots.txt does that — and in June 2026, Google went further than "hasn't confirmed": it stated outright, in its own generative-AI documentation, that llms.txt has no effect on Search rankings or AI Overviews. The honest assessment: llms.txt is a low-cost signal worth adding if your site is complex or context-heavy, but it is not a GEO priority. Fix rendering, crawlers, and schema first.

What llms.txt is

The convention is simple: a markdown file at your root that describes what your site is, who runs it, and which pages are most important for an AI to understand it. A minimal example:

markdown

# BalochDev

> An AI-first software development studio building products for GCC and international markets.

## Key pages

- [About](https://balochdev.com/about): Who we are and what we build
- [Blog](https://balochdev.com/blog): Technical guides on AI search, GEO, and software engineering
- [Services](https://balochdev.com/services): What we offer
- [Contact](https://balochdev.com/contact): Get in touch

## AI Visibility resources

- [How to Get Your Business Found by AI Search](https://balochdev.com/blog/how-to-get-your-business-found-by-ai-search)
- [5-Minute AI Reachability Check](https://balochdev.com/blog/is-your-website-invisible-to-ai)

A fuller version (llms-full.txt) can include complete article content — useful for AI agents that might summarize your site rather than crawl individual pages.

What llms.txt doesn't do

It does not control crawler access — that's robots.txt. It does not replace schema markup. It does not guarantee any AI engine reads it. It is a convention, not a standard with defined behavior from any major AI company.

The evidence against a citation benefit has gotten stronger, not weaker, since the format launched. SE Ranking's analysis of 300,000 domains found 10.13% adoption with no statistically significant citation-rate correlation — and removing the llms.txt variable from their predictive model actually improved its accuracy, meaning the file added noise rather than signal. A separate Ahrefs analysis of 137,000 domains found that 97% of the roughly 38,000 llms.txt files they located received zero requests at all; among the small number that did get traffic, AI bots accounted for only about 1% of requests — most of it came from SEO audit tools and other non-AI bots simply checking whether the file existed.

And as of June 15, 2026, Google made its position official: its generative-AI documentation now states plainly that llms.txt has no effect, positive or negative, on Search rankings or AI Overviews. Google's own John Mueller has compared it to the keywords meta tag — a well-known relic from an earlier era of SEO that search engines stopped using as a ranking input decades ago.

When it does add value

For complex sites with many content types (product, blog, docs, tools all mixed together), llms.txt gives an AI agent navigating your site a useful map — the equivalent of a human-readable sitemap.xml for content purpose rather than URLs. For AI agents performing tasks (agentic browsing, AI coding assistants pulling in documentation), a clear llms.txt can help them identify the right starting point — this is closer to the format's original intent, which was aimed at developer tools and API documentation, not general GEO.

Think of it as a clean welcome mat: low cost, minor upside, not a strategic priority.

The verdict

Add it if it costs you an hour — it signals intentionality, it's genuinely useful for the developer-tool and agentic-browsing use case it was actually designed for, and its value carries some future optionality if a major provider ever does commit to reading it. But don't add it instead of fixing rendering, crawler access, schema, or content structure. Those are the gates that control whether AI can see and cite you at all. llms.txt is decoration on top of a foundation that has to be solid first — and as of mid-2026, the evidence that it moves citation on its own is not just thin, it's actively been tested and found absent.

Frequently asked questions

Does llms.txt help with Google AI Overviews? No. Google's own June 2026 documentation states it has no effect on Search rankings or AI Overviews.

Is llms.txt worth it for a documentation or developer-tool site? More so than for a general marketing site. This is the use case Jeremy Howard originally designed it for — helping coding agents and API-consuming tools navigate documentation efficiently. Anthropic, Stripe, Cloudflare, Vercel, and Supabase all publish one, and that adopter list skews heavily toward developer infrastructure.

Will llms.txt matter more in the future? Possibly, if agentic AI browsing grows and a major provider formally commits to reading it. No major provider has done so as of 2026. It carries option value at near-zero cost, which is the strongest argument for adding it — not a current, measurable citation benefit.

Sources & further reading