Find what
needs redacting.
Point it at a PDF and it surfaces the sensitive data inside — SSNs, emails, phones, cards, dates and IPs by deterministic pattern, and names, organisations and addresses by AI when you bring a key. Review, then redact — the first step of Detect → Redact → Check.
Upload a PDF, review what was found, and redact it in one step. No data is stored.
01 — Upload PDF
02 — Detection depth
Two tiers. A deterministic regex tier (SSN, phone, email, credit card, IP, date) that always runs free and offline. An optional AI tier that finds person names, organisations, and street addresses — the entities patterns can't reliably catch — using your own API key (paid accounts, no key required, are coming).
Confirm what to remove and scrubzero scrubs the text operators from the content stream before drawing the bar — no hidden layers, nothing recoverable. The same engine you'd call as @liiift-studio/pdf-redact.
Verify the result. Switch to Check mode and confirm the redaction actually holds — no text under the bar, no recoverable prior revision, no metadata leak.