Skip to content

The shield: signpost & toll modes

AI crawlers already consume the open web and pay nothing. The standard answer — blocking them — earns zero. The shield puts your own domain in front of a better choice.

Once your domain is connected (see Connecting your domain), you pick a mode per domain:

Human visitors stream through to your site untouched. AI agents get a machine-readable pointer to your paid catalog instead of free content — “here’s the menu, here’s how to pay” — served on your own domain. Use this when what you’re selling is an API or dataset and the website itself should stay free for people.

Agents pay per page (or buy a pass) to browse the site itself — pay-per-crawl, settled on-chain through the same non-custodial split as everything else. Use this when the content is the product: articles, documentation, listings, structured pages.

Real visitors see your own domain, your normal pages, no interstitials, no cookie-walls from us. Google and Bing search crawlers are never tolled — SEO is structurally safe. Only AI agents meet the wall.

Every shielded domain automatically publishes a machine-readable menu of what you sell:

  • /.well-known/ai-catalog.json — your catalog in a standard, agent-readable format
  • /llms.txt — the same story in the convention agents already look for

…both served on your domain, plus a listing in the Superspice directory that agents crawl. Agents find you, see prices, and pay with no human in the loop.

If Superspice is ever unreachable, your site serves normally — the shield structurally cannot take your site down. The worst failure mode is the world before Superspice: agents browse free for a moment.