Redact DOCX for AI

Redact Microsoft Word files before sending to AI.

Word documents are the legacy default for contracts, HR records, and legal drafts. Most contain PII. Cypherz extracts text via mammoth and tokenizes every detector category before any AI sees a single name or address.

  • 01

    Structure-preserving extraction

    Paragraphs and headers preserved during extraction so context for the LLM stays useful.

  • 02

    Custom name dictionaries

    Combine with per-project custom name lists to catch employee or client names that don't fit common patterns.

Upload via the REST API

curl -X POST https://api.cypherz.app/v1/files \
  -H "authorization: Bearer $CYPHERZ_KEY" \
  -F "file=@employment-agreement.docx"

Common questions

Frequently asked.

What happens to my DOCX file when I upload it?

Cypherz extracts text, detects PII, tokenizes it, and stores the original encrypted (AES-256-GCM, per-project key) for download. You get a tokenized extraction back immediately. Files can be deleted at any time and their encryption key destroyed.

What's the max file size?

25 MB on managed cloud. Configurable up to 100 MB on self-hosted deployments.

Can I download a redacted version of the file?

Yes — GET /v1/files/{id}/redacted returns a downloadable redacted artifact. Text-native formats keep their original shape; binary formats fall back to a clean PDF.

Is the original ever stored unencrypted?

No. From the moment we receive your file, it's encrypted at rest with a project-scoped data encryption key.

Get started

Upload your first DOCX now.

Sign up, create a project, copy your API key. The first request is tokenized in under sixty seconds.