Extract
Extract purchase order data, verified
A structured data extraction API for purchase orders: send a PDF, scan, or photo, get back typed fields, every line item accounted for, and a source reference (page, and text span) for each value, so nothing in the result is a guess you have to trust blindly.
Fields extracted
po_numbercustomer_nameship_toorder_datecurrencytotalTry it now, free
Use the free purchase order to excel tool: no signup, up to 3 pages, results in seconds.
Is this synchronous or do I have to poll?
Documents are queued immediately and processed in the background; poll GET /v1/documents/{id} or register a webhook and get notified the moment it reaches succeeded, partial, needs_review, or failed. There is no in-between state where a document just sits there with no result and no error.
What if the purchase order genuinely has nothing extractable on it?
It's marked `failed` with reason `NO_DATA_EXTRACTED`: pages were processed, but zero fields and zero table rows were found. That's a distinct, visible state, not a result that looks the same as a real purchase order with every field blank.