Replace

Technical

A de-identification method that swaps each detected entity with a numbered placeholder such as <<PERSON1>>, <<EMAIL2>>, or <<SSN3>>. Unlike redaction, the relationships between entities are preserved: every occurrence of the same person, account, or phone number gets the same identifier, so the resulting document is still coherent enough to analyze.

Replace is the right method when a document needs to be shared with an external system (an AI model, a vendor, a researcher) that does not need the underlying values but does need to follow the logic of the document. A complaint that references the same plaintiff fifteen times still reads coherently if every reference is replaced by <<PERSON1>>.

A legend (sometimes called a re-identification map) records the mapping between placeholders and originals so the document can be translated back later. The legend itself is sensitive and must be stored separately under the same controls as the original PII. Replace is technically a form of pseudonymization: the identifiers are obfuscated but the data is reversible by anyone who holds the legend.

Detect & redact sensitive data in your documents

PII Anomalyzer scans PDFs, Word, and Excel files for 55+ entity types using on-device AI. Your data never leaves your machine.