Agentic AI Reference Architecture

Cognition · Software component

State Value Estimator

Software componentCognitionCognition & MemoryVariation point (abstract)arc:StateValueEstimator

An abstract cognition component that estimates the value of a newly expanded search-tree leaf state, producing the reward signal backpropagated through the tree.

Responsibility. Estimates the expected outcome value of a search leaf state.

Also known as: Leaf evaluator, Simulation phase

is invoked byis specialized byis specialized byMCTS Planner: is invoked byMCTS PlannerRollout Simulator: is specialized byRollout SimulatorValue Network Evaluator: is specialized byValue Network Evaluator
Direct neighbourhood (hover for relationship types)

Variants

VariantWhen to choose
Rollout SimulatorChoose when no trained value network is available and simulations are cheap, favouring heuristic rollout policies where domain knowledge exists.
Value Network EvaluatorChoose when training data exists or self-play can generate it and upfront training cost plus per-evaluation inference latency are acceptable.

Relationships

is invoked by dependency

Classification

Patterns
Averaging value prediction with short rollout (AlphaGo hybrid)

Sources

  1. Ch5.5: T. Nguyen, "Monte Carlo Tree Search Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.5. ISBN: 9798244538229.