Documentation
ComplyEdge blocks AI requests and answers that break a rule tied to EU law. Every check is logged: the rule, the law and the time, never the prompt or the AI's answer.
Every rule_id in a response belongs to the open ComplyEdge rule set, which ties each rule to the article and paragraph it applies, and every record names the rule set that decided it (bundle_digest). What a decision record must prove is written down separately, as ADCR, the open Rulebook.
Install the SDK, send your first compliance check, and read the article-cited response when a rule fires.
The REST API: compliance checks with a jurisdiction field, rule metadata, audit-log export, trust surface, and account management.
You keep the text, we keep the proof. Store the event_id next to your own copy, then check the pair with one call when an auditor asks.
Send a prompt through the live engine and read the decision, the rule and its citation, and the event_id of the record.
What one Article 5 block leaves behind, field by field, and how a reviewer checks it without seeing the text.
Embed the live Enforcement Seal and publish your AI Trust Center, both backed by the public trust endpoint.
Five EU AI Act requests with their real verdicts, the rule that fired and the article it cites, or try your own. No login.
Every field in the record, why it exists, who can read it and how long it is kept, for a data protection officer or a works council.
The Article 26 deployer one-pager maps each EU AI Act deployer obligation to the ComplyEdge check or record that supports it, and which duties stay yours. It is the leave-behind for a CCO or counsel.