Get started
How Northdoc works
What Northdoc does with a document, step by step, and where to start: the quickstart, keys, and worked examples.
In short: you send a file, Northdoc reads every word and keeps every picture, and you collect the pages, the pictures and, if you ask, a vector per page. Ask for an extraction or questions and the model adds a tidy answer that says which page each value came from.
-
1 You send a file
A PDF, a scan, a photo or an Office file, or a link to one. Options switch on vectors, extraction and questions; without them you get the pages and pictures.
POST /api/v1/documents -
2 Northdoc reads every page
OCR reads the words and tables. Charts, scans, stamps and signatures are kept as pictures, in their place. Vectors are made if you asked.
status: processing -
3 The model fills in what you asked
Only if you sent an extraction or questions: it reads the words and pictures together, fills in your fields, answers your questions and notes the page and quote for each value.
stage: extracting -
4 You collect the result
Check back until it is done, then read the pages, pictures and vectors, and any data and answers you asked for. Ask more questions whenever you like.
GET /api/v1/documents/{id}