Anonymizer Guide

Transform detected PII using various anonymization methods

After the Analyzer detects PII entities, the Anonymizer transforms them using operators like replace, mask, hash, or encrypt. Choose the right operator based on your use case.

Anonymization Operators

Choose the right operator based on your needs:

Replace

Replaces the entity with a placeholder text. The original value is lost.

John Doe → [PERSON]

Best for: Human-readable anonymized text

Mask

Partially hides the entity while keeping some characters visible for context.

john@email.com → j***@email.com

Best for: Maintaining partial readability

Redact

Completely removes the entity from the text, leaving no trace.

Contact John Doe at → Contact at

Best for: Complete removal of sensitive data

Hash

Creates a one-way cryptographic hash. Same input always produces the same output, but cannot be reversed.

John Doe → a3f2b1c4d5...

Best for: Data analysis while preserving uniqueness

EncryptReversible

Encrypts the entity using AES-256-GCM. Can be reversed using the same encryption key.

John Doe → [ENC:a3f2b1c4...]

Best for: Temporary anonymization with future restoration

Operator Comparison

Operator	Reversible	Preserves Format	Consistent Output	Use Case
Replace	No	No	Yes	Sharing anonymized documents
Mask	No	Partial	Yes	Customer support, logs
Redact	No	No	Yes	Legal documents, FOIA
Hash	No	No	Yes*	Data analysis, deduplication
Encrypt	Yes	No	Per-key	Temporary anonymization

* Hash produces consistent output for the same input value

Using the Anonymizer

Step 1: Analyze First

Before anonymizing, you must analyze your text to detect PII entities. The anonymizer works on the results from the analyzer.

Step 2: Select Entities

Review and select which detected entities you want to anonymize:

Use checkboxes to select/deselect individual entities
Unselected entities will remain unchanged in the output
Click entity type badges to bulk select/deselect by type

Step 3: Choose Operator

Select the anonymization method:

Global operator: Applies the same method to all selected entities
Per-entity operator: Set different methods for different entity types

Step 4: Configure Options

Each operator has configuration options:

Replace Options

new_value - The replacement text (default: [ENTITY_TYPE])

Mask Options

masking_char - Character to use for masking (default: *)
chars_to_mask - Number of characters to mask
from_end - Mask from end instead of beginning

Hash Options

hash_type - Algorithm to use (default: sha256)

Encrypt Options

Requires an encryption key configured in Settings
Uses AES-256-GCM encryption

Step 5: Run Anonymization

Click the Anonymize button
Review the anonymized output
Copy or download the result

Encryption & Deanonymization

The Encrypt operator enables reversible anonymization. You can restore original values using the same encryption key.

Setting Up an Encryption Key

Go to Settings → Security
Under "Encryption Keys," click Add Key
Enter a name and your secret key (32+ characters recommended)
Save the key securely - it's required for deanonymization

Key Security

Your encryption key is stored encrypted in your account. If you lose the key, encrypted data cannot be recovered.

Deanonymizing Text

Go to the Deanonymize tab
Paste text containing encrypted entities (e.g., [ENC:a3f2b1c4...])
Select the encryption key used during anonymization
Click Deanonymize
The original values are restored

Token Costs

Anonymization operations consume tokens based on:

Cost = 1 + 0.2 × ops_count + 0.8 × encrypt_count + 0.1 × entities + 0.2 × text_k

Final = ceil(Cost × 0.5)

Where:

ops_count = number of anonymization operations
encrypt_count = number of encryption operations (higher cost)
entities = number of entities processed
text_k = text length in thousands of characters

Operation	Typical Cost
Anonymize (apply only)	1-5 tokens
Anonymize (full)	2-15 tokens
Deanonymize	1-4 tokens

Best Practices

✅Use Replace for documents that will be shared publicly

✅Use Mask when you need to maintain partial readability

✅Use Hash when analyzing data while preserving entity uniqueness

✅Use Encrypt only when you need to restore original values later

✅Store encryption keys securely - losing them means losing access to encrypted data

✅Test anonymization on a small sample before processing large datasets