Frequently Asked Questions

No. All document processing happens locally on your machine using our local runtime. Your original documents and sensitive data never leave localhost. Only the redacted output (with tokens, not real data) can be shared with AI tools.

No. We have no access to your documents because they never reach our servers. This is by design—we built Redactorr so we literally cannot see your data, even if we wanted to.

Token mappings are stored encrypted on your local machine using AES-256 encryption. The encryption key is derived from your machine's unique identifier, so even if someone copies the files, they can't read them on a different machine.

If mappings are deleted, you cannot reverse the tokenization for those documents. We recommend backing up your ~/.redactorr folder if you need to preserve mappings long-term.

Redactorr's architecture supports compliance because your data never leaves your infrastructure. However, compliance depends on your overall implementation and policies. Consult with your compliance team for your specific situation.

Traditional redaction permanently removes data (e.g., replacing SSN with "XXXXX"). Tokenization replaces it with a reversible token (e.g., "[SSN_a7f2]"). You can later restore the original value using the stored mapping—all locally on your machine.

We achieve 97.38% F1 accuracy across 200,000 validated samples, with 97.70% precision and 97.06% recall. Our test suite covers all 5 major categories (PII, Secrets, Contact, Financial, Healthcare) with 40,000 samples each. See our Testing page for detailed benchmark results.

We detect 50+ data types across 2,910 enterprise-grade patterns including: SSNs, credit cards, emails, phone numbers, API keys, passwords, database connection strings, IP addresses, names, addresses, and more. Pro users can add custom patterns for organization-specific data.

Yes! We extract text from PDFs and apply visual black-box redaction. The redacted PDF shows black boxes over sensitive areas with token labels, ensuring both the visual and text layer are protected.

Our detection uses context-aware analysis to minimize false positives. You can always review the before/after comparison and adjust. Pro users can fine-tune detection with custom patterns.

You need to download and run our local runtime—a small standalone application. No Docker, no complex setup. Just download, run, and open the web dashboard.

macOS (Intel & Apple Silicon), Windows 10/11, and Linux (x64). We provide native binaries for each platform.

Only for initial license verification. After that, the local runtime works offline with a 24-hour grace period. Document processing never requires internet.

The dashboard shows "🔒 Local Runtime Active" when connected. If not connected, it displays "⚠️ Browser Mode" and simulates redaction client-side.

Free users get: All 2,910 detection patterns, PDF support, and 100% local browser processing. Free tier has a 1,000 character limit per document and no AI tokens.

Starter includes: 100K AI tokens/month, Breach Checker, one-shot AI analysis, and all models available. Tokens rollover for 1 month.

Pro includes: 500K AI tokens/month, conversational AI, knowledge base, AI memory, BYOK (bring your own key for unlimited usage), custom patterns, and priority support.

Team (A$99/month + A$33/seat) includes 500K tokens per seat, shared pattern library, SSO & audit logs, with volume discounts. Enterprise offers unlimited tokens, custom deployment, and dedicated support. Contact us for pricing.

Yes! Unused tokens rollover for 1 month. Any top-up tokens you purchase never expire.

Yes. Cancel anytime from your dashboard. You'll retain access until the end of your billing period.