Agentic AI Reference Architecture

Infrastructure · Software component

Workload Controller

Software componentInfrastructureInfrastructureVariation point (abstract)arc:WorkloadController

An abstract orchestrator control loop that continuously maintains the declared replica set of one containerised workload, creating, replacing and updating instances to match desired state.

Responsibility. Reconciles a workload's running replicas with its declared specification.

Also known as: Replica controller

orchestratesdeployed onis configured byis triggered byis specialized byis specialized byAgent Controller: orchestratesAgent ControllerContainer Orchestrator: deployed onContainer OrchestratorDeployment Manifest: is configured byDeployment ManifestContainer Health Prober: is triggered byContainer Health ProberStateful Workload Controller: is specialized byStateful Workload Contro…Stateless Workload Controller: is specialized byStateless Workload Contr…
Direct neighbourhood (hover for relationship types)

Variants

VariantWhen to choose
Stateful Workload ControllerChoose when agents or stores need persistent identity, ordered initialization (e.g., coordinator before workers) or reconnection to specific persistent volumes.
Stateless Workload ControllerChoose for stateless agent replicas that load the same model, keep no local state and execute tasks independently (most worker agents).

Relationships

deployed on structural

is configured by structural

is triggered by dynamic

orchestrates control

Classification

Patterns
Reconciliation loop
Technologies
Kubernetes
Quality attributes
Reliability (ISO/IEC 25010 | NIST AI RMF: valid and reliable)Performance efficiency (ISO/IEC 25010)
Risks mitigated
Replica loss after node failure

Sources

  1. Ch4.3: T. Nguyen, "Container Orchestration and Edge Deployment," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 4.3. ISBN: 9798244538229.