Redact sensitive personal information (PII) from text
Deterministically mask sensitive personally identifiable information (PII) in customer conversations and transcripts. ### Redaction Rules - **Language Preservation**: Output stays in the exact same language as input (`language`). - **Structure Preservation**: Preserves speaker turns, punctuation, whitespace, business context, sentiment, and next actions. - **Standardized Placeholders**: - Personal: `[PERSON_NAME]`, `[DATE_OF_BIRTH]`, `[ADDRESS]` - Contact: `[PHONE_NUMBER]`, `[EMAIL]` - Financial & IDs: `[BVN]`, `[ACCOUNT_NUMBER]`, `[CARD_NUMBER]`, `[TRANSACTION_ID]`, `[GOVERNMENT_ID]`, `[CUSTOMER_ID]`, `[EMPLOYEE_ID]`, `[STUDENT_ID]`, `[POLICY_NUMBER]`, `[MEDICAL_ID]` - **Validation**: Every output is verified against an entity placeholder allowlist to ensure zero hallucinated redactions. - **Streaming (SSE)**: Set `stream: true` to receive progressive redaction via Server-Sent Events.
api/v1/text/redactAuthorization
AuthorizationRequiredBearer <token>
In: header
Request Body
application/jsonRequireddomainAny properties in TextDomain
Optional industry vertical domain for vocabulary specialization. Defaults to 'general conversation' if omitted.
languageRequiredLanguage
Language of the input text. The output text remains in this exact language with all detected PII replaced by standardized bracketed placeholders (e.g. [PERSON_NAME], [PHONE_NUMBER], [EMAIL], [BVN], [ACCOUNT_NUMBER]).
"Afrikaans" | "Amharic" | "Arabic" | "English" | "French" | "German" | "Hausa" | "Igbo" | "Isixhosa" | "Italian" | "Kinyarwanda" | "Sesotho" | "Setswana" | "Shona" | "Spanish" | "Swahili" | "Twi" | "Yoruba" | "Zulu"streamStream
If true, streams output tokens as Server-Sent Events (SSE) text/event-stream. Supported for translation, redaction, and short/detailed summarization. (Not supported for structured summarization).
falsetextRequiredText
Input transcript, dialogue, or document text to process.
1Redaction completed successfully.