Initiate a new Document Trust Analysis to extract data from a given document type.

Related guide: Document Trust Analysis

ℹ️ This endpoint accepts two request body formats — pick the one that matches your use case:

• 📁 multipart/form-data — upload a binary file containing the document to analyse.
• 📝 application/json — reference an existing Workflow Session Applicant by ID.

🔎 Analysis type (analysis_type)

• extraction: true (default): data is extracted from the document. A document type is required. Optionally set fraud_level (standard or advanced) to also run fraud detection.
• extraction: false (fraud-only): no data extraction. Only available with multipart/form-data (direct file upload): type is optional and fraud_level must be standard or advanced. When type is set, the classification checks that the document is of this type (DA_1003 or DA_1004 otherwise). When type is omitted, the classification is only reported in document_classification. Not supported with the application/json (Applicant) variant, where a document type is always required.

On multipart/form-data, analysis_type is sent as form fields (analysis_type[extraction], analysis_type[fraud_level]) and extraction is a string ("true"/"false").

🧾 Tax notice (tax_notice) in France, per fraud_level

The 2D-Doc printed on the French tax notice is read only with standard.

🔓 Endpoint access

• Environments: production, sandbox
• API key scopes: organization, workspace
• Plans: pro, scale
• Add-ons (for production access):
  • For extraction only Document Analysis
  • For Fraud Detection with fraud_level set to standard and Extraction Document Trust Standard
  • For Fraud Detection with fraud_level set to advanced and Extraction Document Trust Advanced