Agentic AI Reference Architecture

Observability & Evaluation · Software component

Change Impact Attributor

Software componentObservability & EvaluationObservability & Evaluationarc:ChangeImpactAttributor

A diagnostic component that overlays deployment and upstream dependency change events on performance timelines to identify which component change most likely caused an observed degradation.

Responsibility. Pinpoints the change (agent version, knowledge base update, dependency) correlated with a regression.

Also known as: Blame assignment, Deployment event overlay, Dependency change analysis

readsreceives data fromreadsTime-Series Metrics Store: readsTime-Series Metrics StoreRollout Manager: receives data fromRollout ManagerAgent Dependency Graph: readsAgent Dependency Graph
Direct neighbourhood (hover for relationship types)

Relationships

reads dependency

receives data from dynamic

Classification

Quality attributes
Maintainability (ISO/IEC 25010)
Risks mitigated
Slow root-cause identification after deployments

Sources

  1. Ch10.4: T. Nguyen, "Human-in-the-Loop," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.4. ISBN: 9798244538229.