No, unless you publish developer documentation. Google Search ignores llms.txt, and no independent study has found a link between publishing one and being cited in AI answers.
Key Takeaways
- Most llms.txt files are never used: Ahrefs studied 137,210 domains. About 38,000 had a valid llms.txt file. In May 2026, 97% of those files got zero requests.
- Google ignores llms.txt for Search: Google says the file does not help or hurt visibility or rankings in Google Search.
- It does not predict AI citations: SE Ranking studied about 300,000 domains. Its model became more accurate when llms.txt was removed as a factor.
- It can help developer tools: llms.txt can be useful when coding agents read developer docs directly. That may matter if you offer an API, bioinformatics platform, or LIMS integration. It is much less useful if you sell reagents.
- Spend your AI visibility budget elsewhere: Focus on content that search engines and AI tools can read, strong scientific proof, and mentions from trusted third parties.
The llms.txt Question Every Biotech CMO Is Getting Asked
Your VP of Sales sends you a LinkedIn post that says sites without llms.txt are “invisible to AI.” A few days later, an agency offers a $12,000 “AI Search Readiness” project and puts llms.txt first on the list. Then, your web developer asks if the file needs to go live before the site launches. The problem is simple: no one can explain what the file does.
This kind of thing happens a lot in life science marketing. A technical idea sounds believable, spreads fast, and gets budget before anyone checks the data.
With llms.txt, the data is now clear. It does not improve AI search visibility. That conclusion is backed by two large studies and a direct statement from Google. For most biotech sites, it should not be a priority.
What is an llms.txt file?
In September 2024, Jeremy Howard of Answer.AI proposed a simple convention: a Markdown file at the root of your domain, /llms.txt, offering a clean map of your site’s most important content, similar to the robots.txt framework. A companion file containing the full text, /llms-full.txt, became a de facto standard later, popularized by documentation platforms rather than by the original spec.
The logic was reasonable. HTML pages are heavy with navigation, scripts, cookie banners, and layout markup, and a model working with a limited context window has to wade through all of it to find the three paragraphs that matter. A pre-cleaned Markdown summary saves that effort.
The marketing world heard something different. Within months, llms.txt was being described as “the XML sitemap for AI,” a discovery file that would tell ChatGPT, Perplexity, and Google’s AI Overviews which pages to cite. That interpretation was never part of the original proposal, and it turned out not to be true.
What the Research Found
Almost no one is requesting the llms.txt file from your website.
Ahrefs studied 137k domains that got traffic in May 2026. About 28% had a valid llms.txt file. Of those files, 97% were not requested once in that month.
Most traffic did not come from LLMs. The few files that did get requests were from SEO audit tools that made up the largest group at 21.7%. Ahrefs says its own crawlers made up about half of that group. All AI-related traffic combined was about 20%. AI retrieval bots, the group most likely to support a live answer, made up just 1.1%.
Ahrefs also noted a limit in its data. Its users tend to be more technical and more SEO-aware than the average website owner. So, the 28% adoption rate may be higher than the web as a whole. Even in that group, though, almost no one was reading the files.
Across 300,000 domains, llms.txt predicted nothing
SE Ranking looked at about 300,000 domains. Only 10.13% had llms.txt. The team used correlation tests and an XGBoost model to see if the file predicted how often a site was cited in AI answers.
In fact, the model became more accurate when llms.txt was removed. SE Ranking summed it up this way: llms.txt “doesn’t seem to directly impact AI citation frequency. At least not yet.”
Adoption was not a strong signal, either. High-traffic sites used llms.txt slightly less often than mid-tier sites, 8.27% versus 10.54%. That looks more like testing than a settled best practice.
Google said it in plain language
In May 2026, Google published an AI optimization guide that addressed this topic. It added a note about llms.txt in June and refined the wording in July. Google’s current position is clear:
“It’s completely fine if you decide to create and maintain LLMS.txt files (or other similar files) for other services or systems that use these files. Doing so will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them.”
Google Search Advocate John Mueller made a similar point. He said no AI service had claimed to use llms.txt, and server logs showed little interest in the file. He compared it with the old <meta name=”keywords”> tag: a self-reported signal that search engines learned not to trust.
Why llms.txt Looked More Important Than It Was
Two things made llms.txt look more important than it was.
- The Lighthouse audit: In May 2026, Chrome Lighthouse 13.3.0 began checking for llms.txt in its default “Agentic Browsing” audit. Because Lighthouse is a Google tool, some site owners assumed the file was needed for rankings. The check is about machine readability for AI agents. A missing file is marked “Not Applicable,” not failed.
- Adoption looked like proof: Docs platforms such as Mintlify, GitBook, and Fern began creating llms.txt files by default. That made it look like OpenAI, Perplexity, and Anthropic had chosen to adopt the format. In many cases, the file simply came from the docs platform. Having an llms.txt file does not mean a company uses llms.txt files from other websites.
Where llms.txt Can Be Useful
llms.txt is most useful when a person or tool asks for it directly. For example, a developer may use Claude Code or Cursor while building against your API (if you have one). The coding agent can load your docs into its context window without waiting for a search crawler. That is different from AI search. It also creates real traffic. Mintlify reported that coding agents made 45.3% of requests across its hosted docs in March 2026. Claude Code and Cursor made up 95.6% of identified agent traffic. Ahrefs saw the same pattern: Claude Code requested llms.txt more often than AI retrieval bots or assistants.
Mueller drew the same line. Google offers Markdown versions of some developer docs because coding assistants can work with them more easily. He also said this is a convenience, not a requirement. Models “can read HTML just fine, so this is imo more of a temporary crutch, perhaps to save some tokens.”
For non-developer sites, Mueller said the file does not make much sense. He also noted that making a Markdown version of product specs is not going to create more sales.
That split fits life science sites well. If you offer an API, bioinformatics pipeline, LIMS or ELN integration, or other software that developers build against, an auto-generated llms.txt file can improve the docs experience. Turn it on if your docs platform supports it.
If you sell antibodies, assay kits, instruments, CRO services, or diagnostics, the case is much weaker. A Markdown copy of your product specs will not, by itself, make ChatGPT cite you.
How Do Life Science Brands Get Cited in AI Answers?
The same way they earn organic rankings. Crawlable content, direct answers to the questions buyers ask, and credible third-party proof. There is no file you can upload that skips those steps.
Scientists are using AI tools to compare vendors and shortlist products or vendors before they ever contact sales. AI answer engines still need web content they can find, read, and cite, which means the work looks a lot like strong SEO:
- Make your science easy to crawl. Put key application notes, protocols, and validation data in indexable HTML. Do not hide all the useful details behind gated PDFs or in website designs that rely heavily on JavaScript.
- Answer real questions. Build pages around the exact questions buyers ask, such as, “What dynamic range should I expect from this assay in serum?” A focused answer is more useful than a broad product page.
- Build trust and authority. Peer-reviewed papers, conference posters, KOL bylines, and trade press coverage help support your claims. Third-party proof matters.
- Fix the technical basics. Make sure search engines can crawl the site. Keep pages fast, use a clear site structure, clear headings, add schema, and maintain a working sitemap.
- Keep claims consistent. Use the same specs and claims on your website, datasheets, and LinkedIn. If those sources disagree, an AI system may surface the wrong version.
The Bottom Line for Your Marketing Plan
An llms.txt file is cheap to publish, and Google says it will not hurt your search visibility. If your docs platform creates one automatically, there is no reason to remove it.
The real issue is time and budget. Money spent on an “AI search readiness” package built around llms.txt is money you cannot spend on better application notes, stronger technical SEO, or fixing product pages that are missing from the index.
If a vendor says llms.txt will make your site visible to AI, ask for server log data that shows that AI systems are requesting the file. The published research says that traffic is very limited.
Want to know how your life science brand shows up in AI search? Our team can review your SEO, content, and AI visibility using your own data. Contact us for an AI search and SEO visibility assessment instead of paying for tactics that only sound technical.
