What is Parallel World: a service that measures and monitors how AI sees your website, for LLMO and AIO

Parallel World measures and monitors how your website looks to AI agents. From training crawlers to shopping agents, it measures how your site looks to AI across 5 stages, and monitors how it changes.

Parallel World measures your site at the 5 stages where AI touches it
  • A. Training
    Who touches your site
    GPTBot, ClaudeBot, Google-Extended, etc.
    What they look at
    robots.txt, Content-Signal
  • B. Search index
    Who touches your site
    Googlebot, Bingbot, OAI-SearchBot, etc.
    What they look at
    Sitemap, structured data
  • C. Fetch at answer time
    Who touches your site
    ChatGPT-User, Claude-User, etc.
    What they look at
    Whether key information is in the raw HTML
  • D. Browsing and actions
    Who touches your site
    Browser agents
    What they look at
    Accessibility tree, forms
  • E. Transactions
    Who touches your site
    Shopping agents
    What they look at
    Whether they stop before payment

Source: the stage table on pw.migurei.ai, read 30 October 2026. It measures how each stage sees your site, not how often AI answers mention you.

What does Parallel World measure?

It does not invent scoring criteria: it gathers the published criteria of Cloudflare, Google, OpenAI, Anthropic and others, plus official tool results where a provider has one, and shows where they conflict and when they change. The aim: when someone asks AI about you, the answer uses the correct information about your company.

How can I check for free whether robots.txt refuses AI bots?

Enter a site's domain name in the AI bot access check (https://pw.migurei.ai/en/check/ai-bots?src=mg-guide). It reads the site's robots.txt once and shows, in a table, which of the main AI bots may fetch the front page (/). The domain you enter is not stored and not logged.

How can I check llms.txt and the sitemap too, for free?

The free scan (free LLMO check, https://pw.migurei.ai/en/free-scan?src=mg-guide) scans one site once and gives you the raw results for its robots.txt, llms.txt and sitemap and for Google PageSpeed Insights. Entering an email address is optional, and a scan can take a few minutes.

How the free scan works: one read-only pass, raw results back
  1. Enter a domainAn email address is optional
  2. One read-only scanNo forms, no logins; it can take a few minutes
  3. Raw results returned as fetchedrobots.txt, llms.txt, sitemap and PageSpeed Insights; no score of our own
  4. Weekly monitoring shows what changedWhere the standards disagree and what moved

Source: the free scan and monitoring descriptions on pw.migurei.ai, read 30 October 2026.

How much does it cost?

Full terms are on the pricing page (https://pw.migurei.ai/en/pricing?src=mg-guide).

What does a scan do to my site?

Scans are read-only. The scanner does not submit forms, add items to a cart, sign up or log in. It does not solve CAPTCHAs or change its User-Agent to get around a block, and it does not access pages disallowed in robots.txt.

The scanner identifies itself with the User-Agent migurei-ParallelWeb/0.1 (+https://migurei.ai/ja/scanner; research@migurei.ai); use the robots.txt token migurei-ParallelWeb to allow or disallow it. Its specification is at https://migurei.ai/en/scanner.

How do Japanese sites treat AI bots?

In Parallel World's survey of the top 1,000 .jp domains on the Tranco list (26 October 2026), 74 sites (10.3%) of the 719 whose robots.txt gave a verdict did not allow GPTBot to crawl the top page. 7 sites (1.0%) disallowed Googlebot, and 105 sites (14.6%) allowed Googlebot while disallowing at least one AI training bot. 29 of the 1,000 (2.9%) had an llms.txt on their own domain.

10.3% of the sites do not allow GPTBot on the top page
GPTBot not allowed (74 of 719 sites)74 / 719 (10.3%)
Googlebot allowed, at least one training bot refused (105 sites)105 / 719 (14.6%)
Googlebot not allowed (7 sites)7 / 719 (1.0%)

Source: Parallel World survey, 26 October 2026, the 719 of the top 1,000 .jp domains on the Tranco list whose robots.txt gave a verdict. robots.txt is a request; it does not show who actually visits.

All the numbers, the method and the aggregate data (CSV) are at https://pw.migurei.ai/en/research/japanese-sites-and-ai-bots?src=mg-guide, updated weekly.

Who runs it?

Parallel World is a service of Migurei (HBLAB). migurei.ai is a subsidiary of 株式会社エイチビーラボジャパン. The seller is 株式会社エイチビーラボジャパン (HBLAB Japan), and invoices are issued by the seller. More about the service: https://pw.migurei.ai/en/?src=mg-guide

What this guide rests on, and its limits

Frequently asked questions

Can I use Parallel World for free?

The free scan scans one site once, for free (https://pw.migurei.ai/en/free-scan?src=mg-guide). Monitoring, which scans one site every week, costs JPY 32,780 per site per month, tax included.

Can it measure how often AI answers mention my company (mention rate)?

No. Parallel World does not track mention rate. It measures and monitors how your website looks to AI crawlers and AI agents, door by door.

Does a scan put load on my site?

At most 50 requests per scan that the scanner sends itself (images and scripts the page loads while rendering are counted under a separate cap), and at least 2 to 3 seconds between requests to the same domain.

Will you do the LLMO or AIO work, or consult on it?

No. Parallel World is a measurement service, not an optimization agency or consultancy.

Published by: Migurei (a subsidiary of 株式会社エイチビーラボジャパン, HBLAB Japan) · Published: · Updated: