If you’ve spent any time in the GEO conversation over the last year, you’ve probably heard someone mention llms.txt. It sounds technical. It has a file extension. People talk about it alongside robots.txt and sitemaps, which makes it feel like developer territory.
It isn’t, really. And if you run a website that relies on being found — especially by the kind of people who now use AI tools to research before they buy — it’s worth understanding.
What llms.txt actually is
llms.txt is a plain text file that lives at the root of your website — the same place as your robots.txt — and tells AI language models how to understand and represent your business.
Where robots.txt says “here’s what you’re allowed to crawl”, llms.txt says “here’s who we are, what we do, and which pages actually matter.” It’s a curated, human-written summary designed to be read by machines — specifically the large language models powering tools like ChatGPT, Perplexity, and Claude.
The format is Markdown. The purpose is clarity. The idea is that instead of letting an AI model piece together an understanding of your business from crawling hundreds of pages — and potentially getting it wrong — you give it a direct, accurate briefing.
Where it came from
llms.txt was proposed in late 2024 by Jeremy Howard, co-founder of fast.ai, as an informal standard for helping AI models navigate websites more effectively. It’s not a W3C standard. It hasn’t been formally adopted by any major search engine. But it’s being implemented by a growing number of software companies, publishers, and professional services businesses — and the AI tools that matter are already paying attention to it.
Think of it the way you’d think of robots.txt in the early days of search: not yet universal, not yet enforced, but a clear signal to the systems that matter about how you want to be treated.
What goes in an llms.txt file
The format is deliberately simple. A well-structured llms.txt file contains:
A brief description of the business — who you are, what you do, who you serve, where you’re based. Two or three sentences maximum. This is the AI’s first impression of your organisation.
Links to your most important pages — with brief annotations explaining what each one covers. Not a sitemap. A curated list of the pages that best represent your work.
Optional: links to key content — documentation, FAQs, articles, or course material that helps an AI understand your expertise and approach.
Here’s a simplified example of what a llms.txt file looks like in practice:
# Warren Groom
> Freelance WordPress designer and developer based in Toronto,
> partnering with agencies worldwide since 2008. I build
> custom WordPress sites — hand-coded, no page builders —
> for marketing, PR, and branding agencies, and their clients.
## Services
- [WordPress Website Design & Build](/websites/): Custom
theme development, ACF builds, white-label delivery for agencies
- [Hosting & Care](/hosting-care/): Managed WordPress hosting
with proactive security, updates, and performance monitoring
- [GEO & AI Visibility](/geo-ai/): Structured data, llms.txt,
and AI visibility strategy for WordPress sites
- [SEO](/seo/): Technical and on-page SEO baked into every build
- [CRO](/cro/): Conversion rate optimisation for WordPress projects
## For agencies
- [Agency Partnership](/agencies/): How I work with agencies —
white-label, client-facing, or behind the scenes
- [Portfolio](/portfolio/): Selected client work
## Learn
- [On-Page SEO Course](/learn/on-page-seo/): Free, self-paced
micro-learning course — 27 lessons across 9 modules
That’s it. No code. No plugin. A text file with a clear structure.
Does your site actually need one?
Honestly — it depends on what you’re trying to achieve.
You probably want one if:
Your business relies on being discovered through research. B2B companies, professional services, agencies, SaaS products — anyone where the decision-making process involves asking questions and comparing options. If someone asks an AI “who are the best white-label WordPress developers for agencies?” and you want a legitimate shot at appearing in that answer, llms.txt is one of the signals that helps.
You want control over how AI represents you. Without llms.txt, AI models construct their understanding of your business from whatever they can crawl and infer. That process is imperfect. A client’s outdated review, an old blog post, a cached version of a page you’ve since rewritten — all of it contributes. llms.txt gives you a direct line to correct the record.
You’re in a competitive space where trust matters. Being cited accurately — with the right services attributed, the right expertise described, the right tone — is a brand asset. llms.txt is a lightweight way to protect it.
You can probably wait if:
Your site is a simple brochure for a local business with no digital discovery ambitions. If people find you through word of mouth and direct referrals, the urgency is low. But even then, it takes twenty minutes to set up and costs nothing — so “probably wait” isn’t a strong argument against doing it.
How to add llms.txt to a WordPress site
This is the part where people expect it to get complicated. It doesn’t.
Step 1: Write the file
Open a plain text editor. Write your llms.txt content in Markdown (see the example above as a starting point). Keep the description honest and specific. Keep the links curated — ten well-chosen pages are better than fifty that pad the list.
Step 2: Save it as llms.txt
Exactly that filename, lowercase, with the .txt extension.
Step 3: Upload it to your WordPress root directory
Connect to your site via FTP or your host’s file manager. Upload llms.txt to the same directory as your wp-config.php file — the root of your WordPress installation, not the /wp-content/ folder.
Step 4: Verify it’s accessible
Visit yoursite.com/llms.txt in a browser. If you can read the file, it’s working. If you get a 404, check the upload location.
That’s the full process. No plugin required, no developer needed, no ongoing maintenance beyond updating it when your services or key pages change.
How llms.txt fits alongside other GEO signals
llms.txt is one signal, not a complete strategy. It works best as part of a broader set of AI visibility foundations:
Schema markup establishes your entity data in a machine-readable format embedded in your page code. Where llms.txt is a summary document, schema is granular structured data at the page level. Both matter.
E-E-A-T content signals — author attribution, sourced claims, demonstrated expertise — tell AI models that your site is a credible source worth citing. llms.txt tells them what to cite you for; content quality tells them whether to trust the citation.
Conversational content structure — pages that directly answer questions, with clear headings and unambiguous introductions — makes your content easier for AI to extract and quote accurately.
llms.txt is the quickest of these to implement. If you’re starting your GEO journey and want a first step that takes under an hour and signals to AI models that you’re paying attention, this is it.
If you’d like to understand more about what GEO is and why it matters before diving into implementation, I’ve covered the basics in a separate post. And if you’d like help implementing llms.txt, structured data, and AI visibility strategy on a WordPress site, that’s something I can help with directly.
