Enter your website. We build one page about your factory from what your own site already publishes.
You check it and confirm it. The page then has a public URL and a JSON endpoint: a buyer who hands that URL to an AI gets what you confirmed, and nothing else.
We are not affiliated with OpenAI or Anthropic. A buyer's AI reads this site through public endpoints, and an ordinary question in a chat app does not necessarily come here.
The domain is enough. We read only pages your own server shows to anyone. This step does not build the page; we run the draft and hand it to you.
This is a static page. The form sends nothing anywhere and builds nothing by itself. Hand the domain to us in person. The inbox is being set up; the address goes here when it works.
Confirmed factories
0 factories
The first factories are planned for this week. The 0 above is the real number of factories that have confirmed, not a page that failed to load.
No factory is live here yet. Yours can be the first.
No sample companies, no placeholder rows. Nothing appears in this table before the factory confirms it.
We read only pages your own server shows to anyone, at most 12 of them, obeying robots.txt, and build a draft.
Every field is marked "from your website, unconfirmed". Categories and processes are a keyword guess and often wrong, so read each line and strike out what is wrong. Confirm in person: we come to the factory, you show the business licence and check this page with us. Confirming by email (verify@factory-lookup.ce-net.com) is not open yet; that address receives nothing until it is.
Your page then has a public URL and a JSON endpoint. Paste the URL into Claude or ChatGPT and it reads that page and answers from it. Search engines and chat tools finding this site on their own takes time; the site is new and they do not find it yet.
This part is not built yet. What follows is the rule it will run under, written down now so you can say no to it now.
A buyer's agent answers only from what you published, and never quotes a price for you.
An inquiry has to state quantity, material, tolerance, need-by and destination, and it goes to the mailbox you gave us. No back office to log into.
A person reads every inquiry before it reaches you. The receipt the buyer's agent gets says queued for approval, never sent.
You can change what is published at any time, and you can take the page down at any time.
An unconfirmed draft page opens only through your own private link. It is listed nowhere on this site, and robots.txt plus noindex ask search engines to leave it out. Anyone holding the link can open the page, so do not forward it.
Yes. No fee, no commission, no list for sale. What we build next is the ordering system; that only works if a buyer's AI can find the factory first.
No. Every email in the verification flow travels from you to us. We do not send bulk mail.
No promise. We publish factory records at /factories.json and /llms.txt, which a buyer's AI can read. An ordinary question in a chat app does not necessarily come here. We are not affiliated with OpenAI or Anthropic.
A public mailbox (QQ, 163, Gmail) cannot prove who owns the domain, so it cannot confirm a page. Do it in person instead: we come to the factory, you show the business licence, and the record says verified in person.
Then we do it in person. We come to the factory, you bring the business licence, we read the draft to you line by line and strike out what is wrong, and the page publishes after that. On the phone you only have to keep the draft link, or pass it to whoever looks after your website. The page reads on a phone and needs nothing installed.
Only pages your own server shows to anyone, at most 12. Every field carries the URL it came from, the time it was fetched and the SHA-256 of that page, and you can ask to see them.
Change it. After confirmation the value is whatever you say, even where it differs from your website. The source URL and hash stay on the record.
Tell us in person, or send word through whoever handed you the link. At the next publish the record leaves the directory and its JSON file becomes a 404. No exit interview. Once the mailbox is open, one email will do it too.
We read only pages your web server shows to anyone, we obey robots.txt, and we fetch at most 12 pages per domain.
We do not log in, do not buy lists, do not scrape paid company databases, and do not read your mailbox or social accounts.
A published record carries role mailboxes only (sales@, info@, market@). No personal names, no personal mobile numbers, no photographs of people.
Data is stored on Cloudflare, on servers outside mainland China. We do not give you a filing opinion; if your lawyer needs the facts, we hand over the facts.
Measured 2026-09-16: these pages load no third-party script and set no cookie. Cloudflare, as the host, sees a visitor's IP and browser string in its own server logs. We cannot switch that off and we cannot read it.
These files are public and need no login. A buyer's AI can read them directly.