Licence for AI training crawlers
AGMM publishes its guides for people. People, search engines and AI answer engines read them free, as they always have. This page is for AI training crawlers: software that copies pages to train or evaluate a model. A named training crawler that wants a page or a bulk copy is offered a paid licence through x402, an HTTP payment protocol, and the offer states its terms in machine-readable form.
What a payment buys
- Train. You may use the content you paid for to train and fine-tune models.
- Evaluate. You may use it to test and benchmark models.
- Retain. You may keep the copy you paid for, for as long as you keep the models trained on it.
- One payment covers one download of the version that exists at that moment. A bulk copy is bound to the exact file by its SHA-256 hash.
What it does not buy
- No dataset resale or redistribution. You may not sell, publish or share the content, or a dataset made mainly from it, with anyone else.
- No warranty. The content is supplied as it is. AGMM does not promise it is complete, current or fit for a purpose.
- No endorsement. A licence does not let you say AGMM backs your model or your product.
Where the terms and amounts are
Amounts are in the machine-readable offer (the PAYMENT-REQUIRED header and the response body) and in /x402/catalog.json. How the gate treats each kind of visitor is on the policy page, and the developer guide is at /x402/. Text and data mining rights are also stated in /.well-known/tdmrep.json.
An annual licence
If you would rather license the content for a year than pay per download, write to AGMM.ltd@gmail.com. An annual licence is agreed by proposal.