Skip to main content
Extract returns the data requested by your schema. Read the completed result before deciding how your application handles each value.

Read extracted data

Completed Extract responses place document fields at the top level, without a document wrapper. Read values from extracted_data[i].data. The data value retains the shape defined for the extraction, including object, array, scalar, or null. The capture below uses invoice extraction schema. It requests invoice_number as a string and total as a number. The response returns DEMO-001 and 10.0.

Handle missing evidence

An absent field_metadata property means the response supplies no citation metadata for that entry. Do not interpret missing evidence as a confidence score.

Retrieve a pending result

HTTP 202 means the request is still running. Keep the complete Location URL and follow request progress. The endpoint reference includes schemas for pending responses.

Captured response

The demo script produced this HTTP 200 response on September 9, 2026 from a fictional invoice. Example UUIDs replace the job IDs. The response passed the schema check against the saved API specification.