Built to show what machines can actually read

AI Site View is a scanner for the version of your website that machines see. It sends real requests using each AI crawler's published user agent, reads the HTML your server returns, and reports what a machine can establish about your business, with the evidence for every finding.

Published 7 September 2026

---

Why it exists

Two directory websites run by the same studio, 22,143 pages between them, were returning 403 to every AI crawler. Their robots.txt files welcomed all of them. The cause was a bot-protection setting at the CDN, sitting above the origin, and eight free checkers reported both sites as healthy, because every one of them parsed robots.txt and stopped there. AI Site View was built to make the request instead of reading the file, and to report the disagreement when the two do not match.

Who builds it

AI Site View is a Shika Group property. Shika Group is a United Kingdom studio that builds directories, tools and client websites, and the tool grew out of a problem found on the studio's own sites. The people who build it are the people who answer hello@aisiteview.com.

How it measures

The method is public. The score has two components, Technical Access out of 60 and Machine Readability out of 40, and every weight, crawler role and acceptance criterion is written down on the methodology page, which is kept in step with the code by an automated check. Every fact in a report carries one of four labels, tested, extracted, inferred or unknown, and opens into the evidence it was read from. What cannot be established is marked unknown rather than guessed. No language model writes any part of a report.

What we refuse to claim

Nobody outside the companies that run them can tell you what ChatGPT, Claude, Gemini or Copilot will say about your business. We have no privileged access to any assistant, we do not measure recommendation behaviour, and we do not publish per-engine "understanding" figures. We test whether the crawlers can reach your pages and what your pages state, and we label everything by how it was established.

How the articles are written and corrected

Every article in Learn is written by the team that builds the scanner and is checked against live measurement before it is published. Crawler names and user agents are verified against each operator's published documentation and the source is stated. Each article shows its publication date and, when it changes materially, an update date. No article carries a search-volume figure, because we do not have one we can stand behind. If something on this site is wrong, email hello@aisiteview.com with the page and the correction; corrections are made on the page and the update date moves.

Contact

Email hello@aisiteview.com for anything: a question about a result, a repair request, a correction, or a request about your data. How your data is handled is set out on the privacy page.