In April 2024, a researcher published a proposal for a new web standard called llms.txt. The idea was simple: create a convention for a plain text file that websites could publish to give large language models a structured, concise overview of their content.
The file lives at the root of your website – yoursite.com/llms.txt. It uses basic markdown. It contains a brief description of your website, links to your most important pages, and any information you want AI systems to know about you.
Think of it as robots.txt for the age of AI – except instead of telling crawlers where not to go, it tells language models what matters most and what you actually do.
Within months of the proposal, thousands of websites had adopted it. By mid-2025, Perplexity had begun explicitly supporting it. By 2026, it is a clear best practice for any website that cares about appearing in AI-generated answers. Very few maritime websites have one.
Why Maritime Websites Need This More Than Most
AI language models have a specific problem when they encounter maritime websites. It is not that they cannot read them. It is that maritime content is dense with specialist terminology, acronyms, and context that requires domain knowledge to interpret correctly.
Consider a marine survey company’s website. It mentions ISM, ISPS, IACS, flag state requirements, class notation, dry-dock intervals, P&I, H&M, and dozens of other terms that carry specific meaning inside the maritime sector but are ambiguous outside it. An AI reading this content without context has to make inferences. Those inferences are sometimes wrong.
Now imagine that same website has an llms.txt file that opens with:
# Aegean Marine Surveys
> Independent marine surveying firm based in Piraeus, Greece.
> We conduct hull and machinery surveys, pre-purchase surveys,
> damage surveys, and cargo surveys for insurers, owners, and P&I clubs.
> We work with vessels from 500 to 300,000 GT across the Eastern
> Mediterranean and Black Sea.The AI now has unambiguous context before it reads a single service page. It knows this is a surveying firm, their geographic scope, their client types. It can match this company to a query about pre-purchase surveys in the Mediterranean with far greater confidence.
What Goes in an llms.txt File
The specification is deliberately simple. The file uses standard markdown:
1. Title (H1) – the name of your website or company
2. Blockquote summary – a brief, specific description of what the website is and who it serves. This is the most important part.
3. Optional body text – additional context: key facts, credentials, what makes you distinctive
4. Links to key pages – markdown links with brief descriptions
The file should be concise (200–600 words), specific (real maritime terms, real locations, real vessel types), accurate, and plain markdown only. Here is a complete example for a superyacht management company:
# Marlin Yacht Management
> Private yacht management company based in Antibes, France.
> We manage private yachts between 30 and 80 metres for single owners
> and family offices. We do not run charter programmes.
We currently manage six vessels, all flagged under Cayman Islands
or Isle of Man. Our captains hold MCA Master Mariner certificates.
## Key Pages
- [Our approach](/approach/): How we manage vessels and owner relationships
- [Services](/services/): Full scope of management, technical, and crew services
- [Contact](/contact/): How to discuss management of your vesselHow AI Systems Use llms.txt
When Perplexity, ChatGPT in browse mode, or other AI tools retrieve content from your website, they have limited time to process what they find. The llms.txt file functions as a prioritisation signal: it tells the AI which pages contain the most important content and provides a summary that helps it understand the site without reading every page in full.
Direct retrieval queries. When someone asks an AI to research your company specifically, the AI reads your llms.txt first. A clear, accurate file ensures the AI’s summary of you is accurate and complete.
Category queries. When someone asks “who are the best yacht management companies in Antibes,” the AI retrieves multiple websites and synthesises an answer. Sites with clean llms.txt files are easier to process accurately and more likely to appear in the synthesised answer.
Creating Your llms.txt: Step by Step
Step 1: Write your summary. Answer: “If an AI assistant had to describe your business in three to five sentences, what would you want it to say?” Be specific: type of maritime business, services, clients, where you operate. Avoid generic language.
Step 2: List your most important pages – services, team, case studies, FAQ, contact. Add a one-line description to each link.
Step 3: Link to your FAQ section. FAQ content is exactly the kind of specific, direct-answer material that AI tools cite.
Step 4: Upload the file. Save as llms.txt in plain UTF-8 encoding and upload to the root directory of your website. It should be accessible at yourdomain.com/llms.txt with no authentication required.
Step 5: Verify. Visit yourdomain.com/llms.txt in a browser. If you see the plain text file, it is correctly deployed.
Verify It Is Working
A live llms.txt only helps if AI crawlers can actually read it. Four checks confirm the file is doing its job.
Run the first two checks by opening the URL in a browser and viewing the response. If the browser renders your normal site template instead of raw text, the file is being served as HTML and crawlers will read it as a web page rather than a machine index. Fix the server rule so /llms.txt returns the plain file.
The file is uploaded to a subfolder or the media library instead of the site root. AI crawlers only check yourdomain.com/llms.txt. A file at /wp-content/uploads/llms.txt or /docs/llms.txt is invisible. Confirm the exact root URL resolves before assuming the work is done.
There is no central dashboard that reports llms.txt reads yet, so verification is manual. Re-run these four checks after any site migration, hosting change, or redesign, because each of those can quietly move or overwrite the file.
The Broader AI Visibility Picture
llms.txt is one part of a broader shift. To see how your maritime site is currently performing across all of these signals, run the free Maritime AI Visibility Audit. llms.txt is one part of a broader shift. The other elements: Schema.org structured data, content specificity, technical fundamentals (SSL, fast load times, clean robots.txt), and consistent entity signals across your website, LinkedIn, and directory listings. llms.txt does not replace any of these – it complements both by giving AI a structured entry point to your site.
The Timing Advantage
The maritime sector is lagging in adoption by at least two years compared to technology and professional services sectors. That lag is an opportunity. Every maritime website that implements llms.txt now builds a structural advantage that becomes harder to replicate as the practice becomes standard. The file takes under an hour to write and upload. The benefit accumulates for years.
