Agentic AI Reference Architecture

Cognition · Software component

Information Gain Scorer

Software componentCognitionCognition & Memoryarc:InformationGainScorer

A reasoning-step scorer that estimates, via conditional predictive value, how much each step reduces uncertainty about the final answer beyond prior steps.

Responsibility. Measures each reasoning step's informativeness.

Also known as: Informativeness scorer, Conditional PVI scorer

sends data tois invoked byreceives data fromReasoning Quality Scorer: sends data toReasoning Quality ScorerStepwise Reasoning Verifier: is invoked byStepwise Reasoning Verif…Reasoning Chain Decomposer: receives data fromReasoning Chain Decomposer
Direct neighbourhood (hover for relationship types)

Relationships

is invoked by dependency

receives data from dynamic

sends data to dynamic

Quantitative guidance

As stated by the sources; verify before use.

Classification

Patterns
Conditional Predictive Value Index (PVI)Informativeness = mean information gain per stepRedundant-step pruning
Risks mitigated
Verbose, redundant reasoning tracesCircular reasoning

Sources

  1. Ch3.9: T. Nguyen, "Reasoning Quality," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.9. ISBN: 9798244538229.