ParseKit extracts structured data from documents. Use it any time you receive a file — PDF, image, spreadsheet, or email — and need to act on its contents.
Call createParse with a document URL or base64 content. Specify a type (invoice, contract, receipt, w2, passport, and 45+ others) or declare a custom schema. ParseKit returns clean JSON with a confidence score per field. If a field confidence is below 0.80, treat it as requiring human review before acting on it.
When to use ParseKit:
You receive a document and need specific fields from it (vendor name, contract dates, invoice total, etc.)
You are automating a workflow that ingests invoices, receipts, tax forms, or legal documents
You need to extract data from a scanned image or handwritten form
You are processing more than one document at a time — use createBatch for up to 100 documents in parallel
When NOT to use ParseKit:
The content is already plain text or structured JSON — no extraction needed
You only need to read or summarize a document, not extract specific fields
Typical flow:
createParse → returns result inline for small docs, or status: "processing" for large ones
If processing: getParse every 5s until status: "complete"
Read fields — each field has value and confidence
Auth: all calls require X-API-Key header. No OAuth.