Loading
Every page read costs money, so this one sits behind a code while it is in beta. Ask Jayden if you need one.
Read invoices, quotes, dockets and statements of almost any file type and return their line items as clean JSON.
Drop in whatever the supplier sent: a photo of a docket, a scanned PDF, a Word invoice, a spreadsheet of charges. Every line comes back as JSON with the fields that document actually used, and everything else on the page is kept as metadata rather than thrown away.
Large PDFs are split into page ranges and big photos are re-encoded before anything is sent, so a forty-page statement and a twelve-megapixel phone photo both go through. Every setting that changes the result is exposed, remembered and echoed back in the answer, which is what makes a good run repeatable.