SameDayDesk guide

AI readiness audit tools for websites: what a checker can and cannot prove

An AI readiness audit tool checks public access signals that the answer engines document, mainly robots.txt tokens and whether a URL returns text a person can read. It does not prove that ChatGPT, Perplexity, or Google AI Overviews will cite the site. OpenAI says a site opted out of OAI-SearchBot will not be shown in ChatGPT search answers. Google says AI Overviews have no extra technical requirements beyond ordinary Search eligibility, and that no third-party tool has access to Google's internal ranking or AI systems.

Last checked: 2026-09-23.

What each kind of tool is allowed to claim

Tool	What it can check	What it cannot prove
robots.txt reader	Whether the file names OAI-SearchBot, PerplexityBot, or Googlebot, and what Allow or Disallow rules sit in that group.	That the engine fetched the page, indexed it, or cited it. RFC 9309 says robots.txt rules are not access authorization.
URL fetch	The HTTP status and the HTML your server returned to that client.	How a model ranks that HTML. OpenAI, Anthropic, and Perplexity do not publish a score for it.
Google's own report	Impressions in Google's generative AI features, in the Search Console report Google documents.	ChatGPT or Perplexity citations. Google says third-party tools do not have its internal metrics.
Done-for-you audit	A person can run named prompts and record what the live answers say, then point at the gaps above.	A future ranking. Indexing and serving are not guaranteed.

The controls the vendors actually publish

Use the operator's page, not a checklist that treats every bot token as the same switch.

• OpenAI's crawler overview, checked 2026-09-23: OAI-SearchBot is the robots.txt control for ChatGPT search answers. GPTBot is the training crawler. The two settings are independent. ChatGPT-User is for user-triggered visits, robots.txt may not apply, and OpenAI says it is not used to decide Search appearance. A robots.txt change can take about 24 hours to be reflected.
• Perplexity's crawler docs: PerplexityBot is for surfacing sites in Perplexity search results and is not a foundation-model training crawl. Perplexity-User is for user actions and generally ignores robots.txt. The help center article last modified 2026-09-14 says a Disallow stops PerplexityBot from indexing the page text, while a domain, headline, and short factual summary may still be indexed.
• Google's AI features guide, last updated 2025-12-10: a supporting link in AI Overviews or AI Mode must be indexed and eligible to show a snippet. There are no additional technical requirements. The robots.txt control for Search is Googlebot. Google's crawler list, last updated 2026-07-14, says Google-Extended does not affect inclusion in Google Search and is not a ranking signal.
• Google's generative AI optimization guide, last updated 2026-07-10: Google Search does not use llms.txt or other special AI text files. Keeping one neither helps nor hurts Google Search. Structured data is not required for generative AI search, and there is no special schema.org type to add. The same page says no third-party tool has access to Google's internal ranking or AI systems, and it points site owners to the Search Console generative AI performance report.

How to read a free checker

SameDayDesk publishes a free check at /tools/ai-readiness. Treat its output as a reading of public files: crawler tokens, titles, and markup it can see in the response. Do not treat a score as a citation forecast. Google's 2026-07-10 guide is explicit that outside tools do not see Google's AI systems. OpenAI does not publish a readiness score at all.

A checker that only looks at GPTBot is checking the training opt-out, not ChatGPT search. A checker that weights llms.txt as a ranking factor is contradicting Google's guide. A checker that says JSON-LD is required for AI Overviews is contradicting the same guide.

What to do with the result

1. Open robots.txt and compare the OAI-SearchBot, PerplexityBot, and Googlebot groups with the pages linked above.
2. Fetch the URL and confirm the answer is in the HTML, not only in a script. Google's guide says it can process JavaScript when it is not blocked, and that JavaScript sites are harder to debug. Put the answer in text anyway so a reader, and any fetcher, can see it without guessing.
3. For Google, use Search Console, including the generative AI report Google documents. Do not buy a tool that claims a private Google rank.
4. If you need someone to run the live prompts and write down who gets named, that is a different job from a URL paste. Use the audit linked below. This page does not set its price.