llms.txt is not the AEO silver bullet it was marketed as. But it's also not useless. The truth is more nuanced — and understanding the nuance will save you from wasting time on the wrong technical fix.
What llms.txt Was Supposed to Be
When Anthropic's Jeremy Howard proposed the llms.txt standard in September 2024, the idea was appealing: a single file in your site's root directory that tells AI models what content exists on your site, how it's structured, and what the AI should prioritise. A curated sitemap designed specifically for LLMs rather than traditional search crawlers.
What the Evidence Actually Shows
The honest picture is more limited:
- •No confirmed citation benefit — There is no peer-reviewed or large-scale study demonstrating that adding an llms.txt file directly improves citation rates from ChatGPT, Perplexity, Gemini, or other AI platforms.
- •No confirmed platform adoption — As of early 2026, none of the major AI platforms have publicly confirmed that their crawlers or models use llms.txt files in a standardised, documented way.
- •Real value for developer tools — Where llms.txt demonstrably helps is in a different use case entirely: helping AI coding agents like Cursor, Claude Code, and GitHub Copilot understand how to work with your API or SDK.
- •Markdown copies don't help — A related trend — creating markdown copies of every page on your site — adds duplicate content without a clear benefit.
What You Should Do Instead
The time spent implementing llms.txt is better spent on technical fixes with direct, evidence-backed impact:
- •Fix JavaScript rendering issues — this has a confirmed, measurable effect on AI crawlability
- •Add JSON-LD schema markup — structured data directly improves AI's ability to understand your content type and context
- •Submit an accurate sitemap.xml to all major search engines
- •Review your robots.txt and firewall rules for AI bot user agents
- •Improve your Bing indexation — ChatGPT's browsing capability relies heavily on the Bing index
The Bottom Line
If you've already built llms.txt for your developer documentation, keep it — it's doing something useful there. If you've built it because you believed it would improve your ChatGPT visibility, the evidence doesn't support that belief yet. Focus your technical effort on the five fixes above first.
Hema's 6-module Site Health audit covers all the technical fixes with confirmed impact: rendering, schema, firewall rules, sitemap, content depth, and data tags. These are the issues that actually move your AI visibility score.