SEO · 10 min read · Updated:
What is llms.txt? A guide to AI SEO (GEO) and showing up in ChatGPT
Your customers now ask their questions not only on Google but also to AI assistants. llms.txt is a small, easy step in this new picture; the real work is still in your content.
In short
- llms.txt is a plain text file at the root of a site that describes the site and its key pages to AI tools in Markdown.
- llms.txt is a practice proposed in 2024; it isn't an official standard, and Google says no extra file is needed to appear in its AI features.
- GEO (generative engine optimisation) is the work of getting your site used as a source in AI answers, and its foundation is good SEO.
- Blocking the search bots of tools like ChatGPT, Claude and Perplexity in robots.txt stops your content from being a source in their answers.
- Heyşirket sites generate llms.txt automatically; the AI visibility screen in the Professional plan shows bot crawls, AI-referred visits and a readiness score.
“Which accounting software suits a small business?” Today many people ask this to an AI assistant instead of Google. The assistant builds its answer from pages on the web. Is your page among those sources?
In this article we answer what llms.txt is, what it does and how to write one. Then we look at the bigger picture: what GEO, or AI SEO, means, and what to do in robots.txt, structured data and content to show up in ChatGPT and similar tools. No hype: we separate what is proven from what is still only a proposal.
Part 1 · Definition What is llms.txt?
llms.txt is a plain text file at the root of your site, that is at examplefirm.com/llms.txt. It tells large language models (LLMs) and the AI tools that use them what your site is, who you serve and which pages matter most, in a short and clean format.
The idea was published as a proposal at llmstxt.org in 2024. The reasoning: a web page has menus, ads, cookie banners and code, so it's hard for an AI tool to extract the real information. llms.txt gives the site's summary without the noise.
Part 2 · Format What does an llms.txt file look like? An example
The file is written in Markdown. The proposed structure has four parts: a single heading with the site name at the top, a short summary starting with “>”, then sections opened with “##”, each with described page links.
| Line | Example | Purpose |
|---|---|---|
| # Heading | # Example Physiotherapy | Name of the site or business |
| > Summary | > A centre in Konak, Izmir offering physiotherapy and exercise programmes. | What you do and where, in a sentence or two |
| ## Section | ## Services | Groups pages by topic |
| - name: description | - Back pain programme: session flow and length | Link to a key page with a one-line description |
Some sites also publish llms-full.txt, which collects the full text of their pages in one file. As an example, see heysirket.com's own files: heysirket.com/llms.txt gives the summary and page list, llms-full.txt gives the detailed text, and the English version is at /en/llms.txt.
Writing an llms.txt for your own site
Write the summary
In two sentences, say what your business does, who it serves and where it works. Clear facts, not marketing language.
Pick the key pages
Service pages, the pricing or plans page, contact, FAQ and your most useful guides are enough. You don't have to list every page.
Add one sentence per link
Say what question the page answers. The assistant uses this description to match the page to the user's question.
Put it at the root and keep it current
Publish the file at the root of the site. Update it when you add a new service or an important page.
Part 3 · robots.txt What is the difference between llms.txt and robots.txt?
Both files sit at the root of the site and their names look alike, but they do different jobs. robots.txt is a permissions file: it says which bot may crawl which part of the site. Search engines and the bots of major AI companies follow it. llms.txt is an introduction file: it doesn't allow or forbid anything, it only offers a summary of the site.
| robots.txt | llms.txt | |
|---|---|---|
| Job | Crawl permissions and blocks | Summary of the site and its key pages |
| Status | An established, widely accepted rule | An unofficial practice proposed in 2024 |
| Effect | One wrong line can remove the site from search entirely | Unproven effect, harmless |
| Priority | Get this right first | Can be added afterwards |
In practice, check robots.txt first for AI visibility. A single “Disallow: /” line or an accidentally blocked bot has a bigger effect than anything llms.txt could offer. To see your own file, just add /robots.txt to the end of your site's address in the browser.
Part 4 · GEO What is GEO? What does AI SEO mean?
GEO stands for generative engine optimisation, often simply called AI SEO. Its goal is to get your site used as a source in the answers of tools like ChatGPT, Gemini, Claude and Perplexity, and in Google's AI Overviews.
GEO doesn't replace SEO. Most of these tools search the web and crawl pages when they answer. A page that can't be crawled, loads slowly or doesn't answer the question clearly falls behind in both classic search and AI answers. We cover the basics in what is SEO.
Part 5 · In practice How to show up in ChatGPT and AI search
There's no fixed formula, and no one can guarantee an assistant will recommend you. But the things that raise a page's chance of being chosen as a source are clear:
- Keep the door open to bots: ChatGPT's search feature reaches pages with OAI-SearchBot and ChatGPT-User, Claude with Claude-SearchBot and Claude-User, Perplexity with PerplexityBot. If your robots.txt blocks them, your content can't be a source in those answers.
- Answer the question in the first sentence: Define the topic clearly in a sentence or two at the top of the page. Assistants tend to quote sentences like these.
- Write verifiable facts: Address, opening hours, service scope, plans and prices should be clear. Vague marketing lines don't get quoted.
- Use structured data: Organisation or local business schema, FAQ and article schema help machines understand your page correctly.
- Answer common questions: Turn the questions customers really ask into headings and give short, complete answers. You can find which questions are searched with keyword research.
- Write titles and descriptions with care: The title and description in search results are among the first things assistants look at when picking a page. Details: how to write a meta description.
- Stay current: Outdated prices or a closed branch hurt trust. Review your key pages regularly.
Search bots and training bots are different
Not all AI bots do the same job. Some read your page while looking for an answer to a user's question; these matter for visibility. Others only collect data to train AI models, such as GPTBot or Google-Extended. Blocking training bots is a matter of preference; blocking search bots means not appearing in those tools.
| Bot | Owner | Job |
|---|---|---|
| Googlebot | Crawling for Google Search and AI Overviews | |
| Bingbot | Microsoft | Crawling for Bing search |
| OAI-SearchBot, ChatGPT-User | OpenAI | ChatGPT search and reading pages on a user's request |
| Claude-SearchBot, Claude-User | Anthropic | Claude's search and reading pages on a user's request |
| PerplexityBot | Perplexity | Crawling for Perplexity search results |
| GPTBot, Google-Extended, CCBot | OpenAI, Google, Common Crawl | Collecting data for model training |
Part 6 · Product llms.txt and AI visibility in Heyşirket
Heyşirket sites generate llms.txt automatically from the site's summary and pages; you don't need to do anything. In the company panel, on the SEO ayarları (SEO settings) screen, the Yapay zekâ araçları (llms.txt) section lets you add extra information to the file, for example your service area or how you work. You can also turn the file off with the option not to publish llms.txt.
The Yapay zekâ eğitimi (AI training) preference on the same screen only affects bots that crawl to train models; it's allowed by default. Search and AI answer bots are always open, so you can't accidentally make yourself invisible with this setting.
The SEO › Yapay zekâ görünürlüğü (AI visibility) screen in the Professional plan shows three things:
- Bot crawls: how often search and AI bots crawl your site and which pages they read most. The counts come from the user agent in request logs; fake requests posing as bots may also be counted.
- AI-referred visits: visitors who came to your site from tools like ChatGPT, Perplexity and Gemini. This data comes from your Google Analytics connection.
- Readiness score: whether the site is open to search engines, AI search bots aren't blocked, organisation or local business schema, FAQ schema, page schemas, the sitemap and an llms.txt check. Because the effect of llms.txt is unproven, it carries little weight in the score.
For plan details see the SEO tools and Pricing pages. Step-by-step in the panel: Help › SEO settings.
AI search readiness checklist
- My robots.txt doesn't block AI search bots.
- There's a current llms.txt at the root of my site.
- Every service page says clearly in its first sentence what it offers.
- Address, hours, phone and service area are clearly written on the site.
- Common questions are answered briefly and completely on my own pages.
- Business and FAQ schema are on the site.
- My prices and plan details are up to date.
What is llms.txt?
llms.txt is a plain text file at a site's root that describes the site and its key pages to AI tools in Markdown. It was proposed in 2024 and is not an official web standard.
Does llms.txt affect SEO?
Google says no extra file is needed to appear in its search and AI features. The effect of llms.txt on being quoted in AI answers is unproven; since it's easy and harmless, it can still be added.
What is GEO?
GEO means generative engine optimisation. It's the work of getting a site used as a source in the answers of AI tools such as ChatGPT, Gemini and Perplexity and in Google's AI Overviews; its foundation is good SEO.
How do I show up in ChatGPT?
No one can guarantee it, but not blocking OAI-SearchBot and ChatGPT-User in robots.txt, writing pages that answer questions clearly, giving verifiable facts and using structured data all raise your chance of being chosen as a source.
Should I block AI bots?
If you block bots that search, your content can't be a source in those tools' answers. Blocking bots that only collect data for model training is a matter of preference and doesn't directly affect search visibility.
Check your AI visibility See which pages bots crawl, AI-referred visits and your readiness score. Open AI visibility
Author: Heyşirket Editorial Team