Feature

Structured Data Extraction

A stable, normalized JSON shape you can code against — whichever engine ran.

What is Structured Data Extraction in Mahad OCR?

Available now. Different engines produce different raw output. Input: Any processed document. Limitation: Normalized keys cover common identity and financial fields.

Available now

What it does

Different engines produce different raw output. Mahad OCR normalizes them into one schema so your integration does not change when a document takes a different path.

How it works

1Upload any document
2Engine result is normalized
3Both normalized values and raw fields are returned

Sample result

Illustrative response shape using synthetic data.

{ "values": { "full_name": "…", "document_number": "…", "expiry_date": "2028-01-01" } }

Supported inputs

Any processed document.

Fields returned

values{} — normalized snake_case keys
fields[] — raw label/value pairs as printed
line_items[] for financial documents
field_status and field_evidence per field

API

Present on every document response.

Security & privacy

No change to isolation or audit.

Current limitations

Normalized keys cover common identity and financial fields. Unusual document-specific fields appear in fields[] but may not have a normalized key.