
GEO service
Crawler access, llms.txt, schema and content: end to end, so you appear in AI answers.
See the GEO serviceMeasure how ChatGPT, Gemini and Perplexity talk about your brand with Ossvy, the tool we built.
ossvy.com · AI visibility measurement
Measure what AI says about you → Ossvy
Crawler access, llms.txt, schema and content: end to end, so you appear in AI answers.
See the GEO service
How to fix every check in this tool, step by step.
Download the e-book
Prepared by: Zehra AdaşManaging Partner
Reviewed by: Semih BayındırSenior Software Engineer & Tech Lead
Last updated
When ChatGPT, Gemini, Claude and Perplexity answer a question, they read the web with their own crawlers. If a page is closed to those crawlers, or what it says is unclear when they read it, it cannot be cited as a source. The AI readiness check measures with eight checks whether the page you enter is ready to be read that way.
AI answers now sit above the search results (Google AI Overviews, ChatGPT search), and people often read only that answer. The first condition for appearing in it is technical: the crawler has to reach the page and understand its content.
However good your content is, a single line in robots.txt can remove your brand from an engine’s answers entirely. That is why the check starts with these settings, the ones most often overlooked.
When you enter a URL, our server reads just three files as RoicoolBot/1.0: the domain’s robots.txt and llms.txt and the HTML of the page you entered. We do not crawl the site or open other pages, and every request has a 10-second timeout.
Each check gets one of three results: a pass earns full points, a warning half, an error none. The checks fall into four areas, and the score out of 100 is weighted by area. 0–49 means “Not ready”, 50–79 “Partly” and 80–100 “Ready”.
Our own site's result: roicool.com scored 97/100 (Ready) in the check of 21 October 2026; 11 of the 12 checks passed and 1 gave a warning. We re-run it and update the date when the site changes.
| Area | Weight | roicool.com · 21 October 2026 |
|---|---|---|
| Crawler access (4 bots + Google-Extended) | 40 | 40 |
| AI files (2, proposals) | 12 | 9 |
| Structured data (1) | 24 | 24 |
| Core SEO (4) | 24 | 24 |
| Total | 100 | 97 |
This check is a snapshot; it does not guarantee that you will appear in AI answers. A ready site may still not be cited: content, authority and competitors decide that too.
The check looks only at the page you enter; other pages on your site may score differently. We update the bot list and the rules as they change; the date is shown above. The score and result are not stored: they are kept in server memory for 1 hour and then deleted. No public page is created per domain either. If you ask for the detailed report, only your email address is stored, to send it.
To measure how your brand actually appears in answers, see Ossvy; to close the gaps step by step, read AI visibility: a guide to measuring and improving it; to close them together with us, see our GEO service.
What people ask most about the AI readiness check.
Allowing them only lets your public pages be read; areas behind a login are not affected. You can manage training and search bots separately: for example, block GPTBot and keep OAI-SearchBot open to appear in ChatGPT search.
No. llms.txt is a proposal (llmstxt.org), not a standard, and no major AI engine has officially said it uses the file. That is why the check gives a warning, not an error, when it is missing. It is easy to add and does no harm.
This check measures only the technical prerequisites: can crawlers reach your site, and can your content be read correctly? Whether an engine recommends you depends on how well your content answers questions, whether your brand appears in third-party sources, and on your competitors. You can measure how you actually appear in answers with Ossvy, the tool we built.
No. The check reads your domain’s robots.txt and llms.txt and the HTML of the page you entered once. Every request has a 10-second timeout; it does not crawl the site or open other pages.
The on-screen result summarizes the eight checks. The detailed report adds a code example and a priority order for every row and arrives in your inbox as a PDF. We use your email address only to send that report, and we ask for separate, optional consent for marketing emails.
The result is never published on a public page. For the same domain, it is kept on the server for 1 hour to speed up repeat requests, and then deleted. We keep your email address only if you ask for the detailed report, to send it. Details are in our Privacy Policy.
Let’s prioritize the gaps together; we will put the GEO scope in writing.