About this agent
Static survivorship rules guess which source is most reliable. This agent measures it. By analyzing historical steward override patterns, source system accuracy rates, and attribute-level confidence metrics, it dynamically selects the best value for each golden record attribute — and improves over time as more steward feedback accumulates. The result: fewer overrides, higher golden record accuracy, and survivorship logic that adapts as your data landscape changes.
Use cases
Dynamic attribute-level survivorship using ML-derived source reliability scores
Learning from historical steward override decisions to improve future selections
Detecting survivorship conflicts and flagging for steward review
Generating source system reliability reports by domain and attribute
Monitoring golden record quality drift and triggering re-survivorship
Features
ML-based source reliability scoring per attribute and domain
Historical steward decision learning and model updates
Attribute-level confidence calculation and conflict detection
Full survivorship decision audit trail with reasoning explanations
Configurable learning rate and override thresholds
Integration with Reltio's native survivorship configuration
Tech stack
PythonReltio APIVertex AIBigQueryCloud Functions