Features

Stop redacting documents by hand. Let AI do the heavy lifting.

PII Anomalyzer uses dual AI models to detect PII, paired with a manual draw-to-redact toolbar, so AI does the first pass, you make the final call, and you never upload a file to do it.

Start Your Free 7-Day Trial Take the Product Tour
Detection

Context-aware AI, not just regex

PII Anomalyzer runs two independent AI models on every scan and combines their results. The dual-model approach catches what single-model tools miss: "Jordan" is recognized as a person in one sentence and a country in the next.

  • Wide range of entity types detected automatically
  • Dual-model detection for higher accuracy
  • Pattern recognizers for US, UK, Canadian, AU, and NZ postal formats
  • Surface form propagation — detect once, match everywhere
  • Form field PII detection for fillable PDFs

Entity Categories

Government IDs

Financial

Healthcare

Digital Presence

Travel & Transportation

Location

Insurance & Compliance

Contact Information

Organization & Work

Personal Identifiers

Assets & Devices

Legal & Sensitive

View all entity types ↓
Entity Reference

Built-in entity types

PII Anomalyzer detects personal identifiers, financial data, healthcare records, government IDs, and more — all using context-aware AI, not just regex.

Government IDs

  • US Social Security Number
  • US Driver's License
  • US Passport Number
  • Passport Expiration Date
  • Identity Card Number
  • Identity Document Number
  • National ID Number
  • Birth Certificate Number
  • Visa Number
  • Tax Identification Number
  • Canadian Social Insurance Number
  • Canadian Business Number

Financial

  • Credit Card Number
  • Credit Card Expiration Date
  • Card Security Code (CVV/CVC)
  • Credit Card Brand
  • Cryptocurrency Wallet
  • Currency Amount
  • IBAN
  • Transaction Number
  • US Bank Account Number
  • US Routing Number
  • SWIFT / BIC Code
  • US ITIN
  • Canadian Bank Account

Healthcare

  • Health Insurance ID
  • Medical Condition
  • Medical License
  • Medication
  • National Health ID
  • UK NHS Number

Digital Presence

  • IP Address
  • MAC Address
  • Web URL
  • Username
  • Social Media Handle
  • Digital Signature

Travel & Transportation

  • Flight Number
  • License Plate
  • Reservation Number
  • Train Ticket Number
  • Vehicle Registration

Location

  • Address
  • Location
  • Postal Code
  • Facility

Insurance & Compliance

  • Insurance Policy Number
  • Insurance Provider
  • Registration Number

Contact Information

  • Email Address
  • Phone Number

Organization & Work

  • Organization
  • Student ID

Personal Identifiers

  • Person Name
  • Date / Time

Assets & Devices

  • Serial Number

Legal & Sensitive

  • Nationality / Religious / Political

Other

  • Generic PII
Your Own Rules

Teach it the identifiers only you know

Case numbers, matter IDs, client codes: every practice has formats no built-in model knows. Describe your own formats, list the names and numbers that must always be caught, and list the terms that should be left alone. Your rules run alongside the AI, or on their own.

  • Custom Patterns for your own formats, each with its own placeholder, like <<CASE_NUMBER1>>
  • Test a pattern against the open document before you add it
  • Deny List entries are always caught, with the type and placeholder you choose
  • Allow List entries are never flagged
  • Import either list from a CSV or Excel file
  • Rules-only mode: switch off automatic detection to run just your lists and patterns

Rule Types

Custom Patterns

Your own formats, with their own label

Deny List

Always catch these names and numbers

Allow List

Never flag these terms

Image Deny List

Black out whole embedded pictures

Supported Formats

PDF

Native text, forms, scanned

DOCX

Word documents

Excel

XLSX, XLS, XLSM, XLSB

Images

JPG, PNG, TIFF

Plain Text

Paste or load directly

Scanned Docs

Built-in OCR (multiple engines)

Document Mode

Import any document, see results side by side

Import a PDF, Word document, Excel spreadsheet, or image. PII Anomalyzer auto-detects the document type — native text, fillable form, or scanned image — and applies the right processing pipeline. A side-by-side viewer shows the original next to the de-identified version with native vector PDF rendering. New to this? Follow the step-by-step guide to redacting a PDF.

  • Auto-detects PDF type: native, form, or scanned
  • Non-PDF formats converted automatically
  • Side-by-side original vs. de-identified view
  • Coordinate-aware text processing
  • Built-in OCR for scanned documents
  • Crooked and sideways scans are straightened before they are read
  • Option to read every page of a PDF, Word, or Excel file as an image
De-identification

Four ways to protect sensitive data

Choose the right method for your workflow — from permanent removal to non-destructive annotation.

Redact

Black boxes permanently cover PII. The underlying text is removed from the PDF layer — gone for good.

Destructive

Replace

Numbered identifiers like <<PERSON1>> replace the original text. A legend page is appended for reference.

Destructive

Highlight

Semi-transparent overlays in entity-specific colors mark detected PII for review — non-destructive.

Non-destructive

Mask

Character-level replacement that preserves text length while hiding the value.

Destructive
Draw-to-Redact

Manual precision for what AI misses

Handwritten text, logos, signatures, and visual PII in images — some content needs a human eye. Draw redaction rectangles directly on the PDF, resize with drag handles, and nudge with arrow keys for pixel-perfect placement.

  • Select, Draw, Delete, and Clear All toolbar modes
  • Click-and-drag to create redaction boxes on any page
  • Resize by dragging edges, nudge with arrow keys
  • Always renders as solid black — explicit "remove this" intent
  • Persists between de-identification runs for iterative refinement
  • Zoom-independent — coordinates stored in PDF points

Toolbar

Select

Click to select, drag to move

Draw

Click and drag to create redaction boxes

Delete

Remove selected rectangle

Clear All

Remove all rectangles

Batch Processing

Process multiple documents at once

Queue multiple documents for processing in a single workflow. Each document gets its own de-identified output, modified PDF, and results table — combined into a single exportable dataset. If a run is interrupted, it recovers cleanly instead of leaving a half-written file behind.

Re-identification & Export

De-identify, share safely, translate back

Export a de-identified document using the Replace method, share it with AI tools or external reviewers, then use re-identification mode to translate their responses back to the original context. One restoration key can restore several related files, and you can also use it to verify redactions before final release. Export de-identified text, results tables (XLSX), and modified PDFs. Remove hidden data, on by default, strips metadata, photo location data, and leftover form-field values from the exported file.

.txt De-identified text
.xlsx Results table, plus the Allow List and Deny List applied
.pdf Modified PDF with redactions applied
Privacy by Design

Fully offline PII detection, redaction, and anonymization.

Every AI model, every algorithm, every computation runs on your machine. No document data leaves your computer. No telemetry, no analytics, no usage data sent back to us. There is no server holding your documents because there is no server.

Offline processing

Detection, redaction, and anonymization work without an internet connection

All models bundled

NLP and OCR engines ship with the app — nothing to download separately

No telemetry

No analytics, no usage data leaves your machine

Try it on a real document

A contract, a scanned record, a spreadsheet of account numbers. The AI finds the PII, you choose what gets redacted, and nothing leaves your machine.

Start Your Free 7-Day Trial View Pricing

7-day free trial · Windows & macOS