Are AI search bots actually looking at llms.txt files?

The idea behind llms.txt is straightforward: give AI crawlers a structured, easily digestible version of your site’s most important content. But does it work?

Yoast recently added the ability to autogenerate llms.txt to their plugin, but John Muller from Google recently posted that no AI system currently uses llms.txt.

When llms.txt first “became a thing” I launched a plugin to auto-generate llms.txt for WordPress thinking it would be an internal tool for evaluation on our client and internal sites, but the plugin has grown organically and as of today, it has 78,764 downloads.

So that brings me to an easy experiment to answer the question: Are AI search bots actually looking at llms.txt files?

The Experiment

Track which AI crawlers actually visit llms.txt when it’s available.

The latest version of the plugin includes an opt-in feature that logs when known AI bots hit the file. If users opt in:

• Their site logs bot visits locally with the name of the bot and the date

• There’s a live view of site-specific logs inside WordPress

• An anonymous signal is sent to our server and that data contributes to a public dashboard

All data is encrypted and anonymous. The data shared includes the bot name, timestamp, and a hashed version of your domain to track LLM crawler behavior across thousands of sites. No content or personal information is collected or stored.

Bot tracking

• GPTBot (OpenAI)

• ClaudeBot / Claude-Web (Anthropic)

• PerplexityBot

• ChatGPT-User

• MistralAI-User

• Bytespider (ByteDance)

• Amazonbot

• Google-Extended

• GoogleOther

• Googlebot (for baseline comparison)

This list may evolve, and my plan is to segment the tracking counter by individual bot.

Tracking counter here: https://www.websitellm.com/

I’ll post an update here and at https://x.com/ryhowww with results.