One operational record for everyorder and shipment.

Document digitizer

Teach it your paperwork once. Review what it reads.

Upload PDFs, Word files, Excel sheets or photographs of paper, or forward PDFs, PNGs and JPEGs to a shipment's own email address. Senda reads what the page itself makes readable — its supported tally layout, and the fields of any layout you teach it once — and when that is not enough, the vision model reads scanned and photographed invoices and packing lists, in English or Korean, and the whole of a scanned tally sheet it could only partly read. Everything it returns shows which engine read it and where on the page it sat. What it cannot read goes to your reviewer, never a guess.

Draw the box once

Drag a rectangle round the invoice number on one sample. Later documents of that layout are read from the same coordinates — no model, no escalation to a metered engine, and no deploy to change it. It works for fields that stay put; a total that moves with the length of a table still goes to review.

Every value says where it came from

Each field records the engine that read it, that engine's version, a confidence, the page number and the box on the page. Click a value in review and the highlight is already there.

Word and Excel keep their pages

Office files are laid out to PDF before anything reads them, so a value in a .docx still has a real page and a real box instead of no provenance at all.

Corrections are recorded, not overwritten

A human edit is stored as its own engine at confidence 1.00, so the record keeps both what the machine read and what a person changed.

Extraction is assistive: values arrive with a confidence and go to a reviewer, who has the final say. Senda does not file anything with a customs authority on your behalf.