Visork is AI-powered Trust & Safety infrastructure for online platforms. It is not a moderation dashboard and not a consumer application — it is a detection layer that sits inside your existing stack. Your backend sends user-generated content to a REST API; the API returns a machine-readable result: a probabilistic score, a verdict against your configured threshold, and an audit-log entry.
The division of responsibility is deliberate: Visork performs detection — your platform stays in control of the decision. What happens to flagged content (allow, block, quarantine, escalate to human review) is your moderation policy, applied by your systems.
The first detection module targets one of the hardest problems in online safety: CSAM in user-uploaded images. Detection is ML-first, so it can flag previously unseen material rather than only content already known to hash databases. Inference, transient storage, and audit logs for API content stay in the EU.
How these docs are organized
Visork is in its pilot phase, and the documentation reflects that honestly: these pages cover the concepts — how access works, what detection returns, how your data is handled, and how billing is structured. The full API reference (endpoint schemas, error codes, integration examples) is shared during pilot scoping, once a platform is verified.