Bring your own document types: from samples to automation
DocWise Team · · 4 min read
Why DocWise starts from a few of your own samples instead of a fixed list of templates, and what happens between upload and an automated pipeline.
Most document AI products ship a list of supported documents. If yours is on the list, great; if it is a supplier's delivery note, a lab's report format or an internal form, you are on your own.
DocWise takes the opposite approach: you bring a few samples of the document you receive, and the system proposes the fields to extract.
1. Upload samples
Three to five representative examples are usually enough to start. Include the variations you actually see — different suppliers, scanned and digital copies.
2. Confirm the suggested fields
DocWise suggests field names, types (text, number, currency, date, enum, tables) and validation rules. You rename, add or remove fields and set which ones are required.
Before publishing you can run the type against your samples and compare results with what you expected.
3. Automate
Once published, every incoming document is classified across your enabled types, extracted and validated. Results above your confidence threshold go straight to your systems; the rest wait in a review queue.