Human-in-the-Loop Overview
Human-in-the-Loop (HITL) enables human verification and corrections to ensure accuracy of data extracted by Document AI processors before it is used in critical business applications.
It provides a workflow and a platform for your own or a partner workforce (humans referred to as labelers in HITL) to review, validate, and correct the data extracted from documents by Document AI processors.
Features
- Confidence threshold filters to limit the number of documents going through HITL.
- Labeler pool management, including task assignments and efficiency analytics by task and by labeler.
- UI cues and features that reduce labeler handling time per document.
- Analytics and metrics by task and by labeler, so you can streamline HITL operations.
Benefits
- Risk mitigation - mitigate financial risks of critical data being incorrect - for example, invoice amounts, billing addresses, loan amounts, etc.
- Simplify Exception Handling - Easily roll out a human review and exception handling workflow.
- Workforce Efficiencies - manage, monitor and improve productivity of workforce managing human review.
- Cost control - control costs of human review with configurable filters.
- Data completeness - ensure extracted data is complete for downstream business applications.