Evidence captures information needed to evaluate and approve governed bundles. It documents decisions, validates model performance, and creates audit trails throughout the governance lifecycle.
Evidence takes many forms:
Manual evidence includes text responses, file uploads and structured selections that practitioners provide during review. These inputs answer questions defined in policy stages.
Automated checks run validation logic without manual intervention. Metrics checks evaluate model performance against thresholds. Scripted checks execute custom code for complex evaluations like fairness assessments.
Monitoring checks connect governance to Domino Model Monitoring (DMM). When drift or data quality alerts trigger, governance creates findings with severity levels and assigns owners for follow-up.
-
Manual evidence: add text responses, files, and artifacts to bundles
-
Automated checks: run metrics and scripted validations
-
Monitoring checks: connect to Domino Model Monitoring