As search continues to evolve in 2026, content teams are experimenting with a new way to rank: LLM-focused pages. They’re spinning up assets that no human visitor is ever meant to read directly — markdown files, ultra-minimal JSON feeds, and full /ai/ versions of existing articles. The reasoning feels intuitive: if you present content in a format that’s easier for AI systems to digest, you’ll earn more mentions in ChatGPT, Perplexity, and Google’s AI Overviews. Cut the ads. Ditch the navigation. Deliver nothing but clean, structured text to bots.
Industry observers like Malte Landwehr have highlighted examples of sites generating .md versions of every article or publishing llms.txt files to steer AI crawlers. Some teams are even maintaining complete shadow libraries of their content just for LLMs.
Google’s John Mueller remains unconvinced. “LLMs have trained on – read and parsed – normal web pages since the beginning,” he noted in a recent Bluesky conversation. “Why would they want to see a page that no user sees?” He drew a sharp analogy: LLM-only pages resemble the old keywords meta tag — technically available for anyone to use, but largely ignored by the very systems they’re supposed to influence.
So is this tactic actually effective, or just the newest SEO mirage?
The emergence of ‘LLM-only’ web pages
The movement is very much underway. Sites in tech, SaaS, and developer documentation are rolling out LLM-oriented content formats. The real issue isn’t whether teams are adopting them, but whether these setups are actually producing the AI citations they’re chasing. Here’s what content and SEO teams are putting into practice.
llms.txt files
A markdown file at your…