What an AI agent reads on 味吉兆
An assistant's crawler fetches the page once, without running JavaScript, and reads the text, the structured data and the links. Below is what it gets from bunbu-an.jp, taken from the copy the scanner kept.
105
characters of text without JavaScript, too few to answer questions from
Organization
structured data for the business
2
of 3 AI crawlers turned away at the edge (GPTBot, ClaudeBot)
The words an agent gets
Title: none · language ja. First 105 of 105 characters:
HOME お料理 ご予約 お食事券 店舗案内 特定商取引法 English 簡体中文 繁体中文 한국어 NEWS ◆料理教室の開催日 10月22日（木）11：30～ お問い合わせは050-3628-1055まで
What it can say about the business
| A business type in JSON-LD | Organization |
| Name | 味吉兆 ぶんぶ庵 ｜ 日本料理を大阪本町堀江にて |
| Telephone | no |
| Address | no |
| Opening hours | no |
| Menu or price list | nothing |
| Cuisine in structured data | no |
| JSON-LD blocks | 1 (BreadcrumbList, ListItem, Organization, WebPage, WebSite) |
The links it would act on
| Link | Role | Where it goes |
|---|---|---|
| ajikitcho.jp/reservation/ | booking | an unrecognised third party |
Everything here comes from fetching the homepage once as a crawler. The report turns it into a score and a list of fixes, with code to paste.