Agentic AI Reference Architecture

Cognition · Software component

Decision Fusion Aggregator

Software componentCognitionCognition & MemoryVariation point (abstract)arc:DecisionFusionAggregator

An abstract cognition component that combines conclusions produced independently and in parallel by different decision paradigms into one unified decision.

Responsibility. Fuses parallel paradigm outputs into a single decision.

Also known as: Fusion layer, Result fusion mechanism

receives data fromreceives data fromreceives data fromis specialized byis specialized byis specialized byRule-Based Decision Engine: receives data fromRule-Based Decision EngineUtility-Based Decision Maker: receives data fromUtility-Based Decision M…Learned-Policy Decision Engine: receives data fromLearned-Policy Decision …Constraint-Gated Decision Fusion: is specialized byConstraint-Gated Decisio…Learned Meta Decision Fusion: is specialized byLearned Meta Decision Fu…Weighted Voting Decision Fusion: is specialized byWeighted Voting Decision…
Direct neighbourhood (hover for relationship types)

Variants

VariantWhen to choose
Constraint-Gated Decision FusionChoose when regulatory or safety rules must never be overridden by learned or utility scores.
Learned Meta Decision FusionChoose when training data showing which combinations produce the best outcomes is available.
Weighted Voting Decision FusionChoose when component accuracies on similar cases are known and redundancy should catch errors through disagreement.

Relationships

receives data from dynamic

Design guidance

Classification

Patterns
Parallel integration architectureTimeout and partial-result handling
Quality attributes
Reliability (ISO/IEC 25010 | NIST AI RMF: valid and reliable)Performance efficiency (ISO/IEC 25010)
Risks mitigated
Single-paradigm blind spotsSingle component failure

Sources

  1. Ch5.13: T. Nguyen, "Hybrid Decision Systems Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.13. ISBN: 9798244538229.