Submit Document for Verification

Submit a document for AI and human verification with integrity hashing

Submits a document to the verification pipeline for automated AI analysis followed by optional human review. The platform extracts content, computes integrity hashes, runs anomaly detection, and produces a verification determination.

Documents are processed asynchronously. The response includes a document_id that you can use to poll for results via the Get Document endpoint, or subscribe to the document.verified webhook event.

The platform supports PDF, images (PNG, JPEG, TIFF), and structured data files (JSON, XML). Maximum file size is 50 MB after base64 encoding.

Authentication

Requires X-API-Key header or Bearer JWT token. Tenant-scoped via X-Tenant-ID.

Request Body

agent_id · string · required

The identifier of the agent submitting the document for verification. Must be registered and active within the tenant. Format: maip-agent:ULID.

document_type · string

Classification of the document. Helps the AI pipeline select the appropriate extraction model. Common values: - contract — legal contracts and agreements - invoice — financial invoices - certificate — certificates, licenses, or credentials - report — compliance or audit reports - identity — identity documents (passport, driver’s license) - other — unclassified documents

file_name · string

Original file name with extension. Used for display and to infer content type when not explicitly set.

document_hash · string

Client-computed SHA-256 hex digest of the original file bytes (before base64 encoding). If provided, the platform verifies the hash matches the uploaded content. Recommended for integrity assurance.

document_b64 · string · required

The document content as a base64-encoded string. Maximum 50 MB after encoding.

Response

document_id · string

Unique identifier for the document verification record. Format: maip-doc:ULID.

status · string

Initial processing status. Always processing on submission.

extraction_hash · string

The platform-computed SHA-256 hex digest of the uploaded document bytes.

created_at · string

ISO 8601 timestamp of when the document was submitted.

Processing Pipeline

1. Upload and Hash

   Document is received, decoded, and SHA-256 hashed. If document_hash was provided, it is validated against the computed hash.

2. AI Extraction

   Content is extracted using OCR and document understanding models. Key fields are identified based on document_type.

3. Anomaly Detection

   AI models scan for signs of tampering, forgery, inconsistencies, and metadata anomalies.

4. AI Verdict

   An automated determination is produced with a confidence score.

5. Human Review (optional)

   If confidence is below threshold or the document type requires it, the document is routed to a human reviewer.

6. Receipt Minted

   A MAIP receipt is minted recording the final verification determination.

The document_b64 field must contain only the base64-encoded document content. Do not include data URI prefixes (e.g., data:application/pdf;base64,).