How it works
Define your fields
Set up the classification fields that match your workflow — relevance, sensitivity, data type, document category, or anything custom. Each field gets its own set of values and descriptions that guide the AI.
Run classification
The AI processes your document set in bulk, reading the full text and metadata of each document. Classification runs in parallel across your documents — thousands of documents, not one at a time.
Review by confidence
Every classification comes with a confidence score. High-confidence results are applied automatically. Low-confidence documents are surfaced for human review — so your team focuses on the decisions that actually need human judgment.
Iterate and refine
Adjust field definitions, re-run on subsets, override individual decisions. Refine your taxonomy as you review — then re-classify with updated definitions.
Built for real review workflows
Custom fields — define any taxonomy that matches your workflow, not a preset list
Confidence scores on every decision, so you know exactly where human review is needed
Bulk processing — classify thousands of documents in a single run, not one at a time
Checkpoint resume: classification picks up where it left off if interrupted
Full audit trail on every classification decision with timestamps and attribution
Override and re-classify individual documents or subsets as your understanding evolves