Extractions

Extractions pull structured data from documents. Point an extraction at a document with a schema (or inline fields), and get back clean JSON.

The extraction object


{
  "id": "e1a2b3c4-...",
  "object": "extraction",
  "document_id": "f47ac10b-...",
  "schema_id": "s1a2b3c4-...",
  "data": {
    "vendor": "Acme Corp",
    "total": 1500.00,
    "date": "2026-01-15"
  },
  "confidence": 0.95,
  "duration_ms": 2340,
  "created": 1710000000
}

Run an extraction


POST /v1/extractions

Provide one of the following to define what to extract:

Parameter	Type	Description
`document_id`	string	Required. The document to extract from
`schema_id`	string	ID of a saved schema
`fields`	array	Inline field definitions (same format as schema fields)
`description`	string	Natural language description (10+ chars, auto-generates fields)

Using a saved schema


curl https://docld.com/api/v1/extractions \
  -H "Authorization: Bearer docld_..." \
  -H "Content-Type: application/json" \
  -d '{
    "document_id": "f47ac10b-...",
    "schema_id": "s1a2b3c4-..."
  }'

Using inline fields


curl https://docld.com/api/v1/extractions \
  -H "Authorization: Bearer docld_..." \
  -H "Content-Type: application/json" \
  -d '{
    "document_id": "f47ac10b-...",
    "fields": [
      {"name": "vendor", "type": "string"},
      {"name": "total", "type": "number"}
    ]
  }'

Using a description


curl https://docld.com/api/v1/extractions \
  -H "Authorization: Bearer docld_..." \
  -H "Content-Type: application/json" \
  -d '{
    "document_id": "f47ac10b-...",
    "description": "Extract the vendor name, invoice number, and total amount"
  }'

Retrieve an extraction


GET /v1/extractions/:id


curl https://docld.com/api/v1/extractions/e1a2b3c4-... \
  -H "Authorization: Bearer docld_..."