The PDF analysis API for developers. POST a file or share a link. Define the fields you need, get vendor, totals, and line items back as JSON, no matter the layout.
HOW IT WORKS
List the fields you want back, like vendor, invoice number, total, or line items. You choose them, the API fills them in, regardless of layout.
POST a file directly, or generate a shareable link. Users upload from any browser, no app needed.
A webhook fires the moment analysis finishes, with every field you defined already filled in.
ONE SCHEMA ENGINE
Define the fields you need. The API returns exactly those, structured as JSON, regardless of the document's layout.
Vendor names, dates, totals, account numbers, anything labeled.
Line items, billing breakdowns, schedules, with full row and column structure.
Filled or empty, checkboxes, radio buttons, signatures.
Paragraphs, clauses, sections, with semantic grouping.
Totals, taxes, percentages, with currency and unit detection.
Page number, region on the page, document structure.
Presence detection, signer name extraction.
Define a field, get it back.
USE CASES
Vendors send invoices in 50 different layouts. Extract vendor, invoice number, totals, and line items.
Employees upload receipts via the link. Extract merchant, date, amount, category. No manual entry.
Claim forms and policy documents. Extract policy number, claim amount, incident date.
Patient charts and lab results. Extract patient ID, diagnosis codes, medications.
Reconciliation in seconds. Extract account number, period, opening and closing balance.
Let users share a document instead of pasting text. No PDF infrastructure to manage.
PRICING
FAQ
OCR gives you raw text. We give you structured data. Define a schema and the API returns the vendor, totals, and line items you need, not just a wall of extracted characters.
Scanned documents, native text PDFs, fillable forms, and mixed documents with both. One extractor handles all of them without template setup.
Yes. Line items, billing breakdowns, and schedules come back with full row and column structure, not flattened text.
Yes. Add a signature_present field to your schema, and the API can also extract the signer name where legible.
Most documents process in 1 to 3 seconds per page. The webhook fires the moment processing completes.
Most customers see 95 to 100 percent accuracy on well-defined fields. Every field includes a confidence score, so you can auto-accept high-confidence results.