Demo

Screenshots

  1. Watchlist Your sensitive values, encrypted on disk. Add the names, addresses, and identifiers you always want caught. Stored locally in the macOS Keychain.
    Total Redact watchlist screen showing personal values to always catch
  2. Quick paste Paste text. Get it back, anonymized. Useful for chats, support tickets, and prompts you are about to send to an LLM.
    Total Redact quick paste mode redacting a block of text
  3. File redaction, step 1 Drop in a file. Analysis runs on your Mac. PDF, Word, Excel, PowerPoint, CSV, HTML, plain text, and email files. Processed entirely on your Mac. No cloud service involved.
    Total Redact analyzing a PDF on macOS
  4. File redaction, step 2 Watchlist matches surface first. Data in your watchlist will always be detected for redaction.
    Total Redact showing watchlist matches found in a document
  5. File redaction, step 3 Review every match. Approve, edit, or skip. Out of the box, Total Redact catches names, physical addresses, social security numbers, phone numbers, email addresses, dates of birth, and credit card numbers. Detection runs through a combination of regex patterns and a local machine learning model.
    Total Redact review screen with detected names, addresses, and other PII
  6. File redaction, step 4 Every occurrence, in context. View all occurrences of potentially sensitive data to help you make a decision.
    Total Redact showing every occurrence of a matched value across a document

What goes in. What comes out.

Side by side comparison of an original document with personal data and the same document after Total Redact has replaced names, dates, numbers, and addresses with x's

You pick the replacement.

Many of us now pass personal information to AI. Replacing real names with xxxxxxxx makes the document hard to discuss with the assistant. Replacing “John Doe” with “Bill Smith” gives it a usable name without giving up the real one. Each field is yours to choose. Keep a friendly stand-in for the ones the model needs to reference, blank out the ones it does not.

The same document, redacted with user-chosen placeholder values: Bill Smith, 01/01/1985, 111-11-1111, (111) 111-1111. Email, credit card, and home address remain redacted with x's