Document Classification
Automatically identify what a document is before it's extracted.
Automatically identify what a document is before it is extracted. Document Classification detects a document's type and category from its content using an LLM classifier. It can run on its own to sort an unlabeled inbox or run before extraction to route each document to the correct reviewer.
Use the classification workflow in the product UI. This page covers capabilities, formats, limits, and results.
In the product UI, this capability is available under Capabilities > Classification.
Supported classification types
Document Classification uses the platform's shared document type registry.
Built-in category list
The upload flow includes a built-in category list that can be combined with any custom document types supplied by the caller.
Single classification
Classify one document asynchronously.
No additional limit specified.
Bulk classification
Classify multiple documents in one asynchronous batch.
500 MB maximum per ZIP and 100 MB maximum per file within a ZIP.
Synchronous classification
Classify one document and return the result immediately.
No additional limit specified.
Custom type restriction
Restrict classification to a caller-supplied list of document types.
Supported through custom_document_types or a URL to a JSON type list.
Multi-document PDF handling
Detect and segment multiple documents within a single PDF.
PDFs with 20 pages or fewer use direct LLM classification; longer PDFs use a homogeneity-check pipeline.
Excel workbook classification
Classify each sheet individually, with optional entity mapping.
No additional limit specified.
Confidence scoring
Each result includes a confidence score from 0.0 to 1.0.
No fixed pass/fail threshold is defined by the classifier.
Search and filter
Search the classification list by ID, type, status, or file type, and filter by status.
No additional limit specified.
Confidence scores do not have a fixed pass/fail threshold in the classifier itself. The score is also used internally to help decide whether a multi-page PDF should go through the homogeneity-check pipeline.
Last updated
Was this helpful?

