Skip to main content

Document types

A document type connects a file to the schema that reads it. New uploads no longer require the user to pick or create a type before processing.

How recognition works​

The processing service reads the PDF content and returns a proposed type, confidence and reason. The workspace then:

  • matches the proposal to a known registry or custom type when possible;
  • creates a custom type when the document does not fit the existing vocabulary;
  • associates or prepares the extraction schema used for the next step.

The document row shows AI detected for this path. Registry documents can instead arrive with a type supplied by their source.

Correcting a type​

Open Change type on the document row when the proposal is wrong. The correction is recorded in the activity trail. Read the document again so the chosen schema is applied.

Managing the vocabulary​

The Extraction workspace section lists the reusable types and schemas. Create or edit them there when you want a controlled schema before a document arrives. Types created automatically can be refined in the same place.