Agentic – Scope

Agent architectures (Single, Swarm, Mesh)

Orchestration (Coordinator, Manager/Worker)

Human-AI interfaces (Approval, Monitoring)

Agent safety (Guardrails, alignment)

Agent observability (Tracing, metrics)

Cost & performance of agent operations

Prompt Engineering: Single prompts addressed in other pillars

LLM Training: Model training is infrastructure/operations

Data Engineering: Training data is out of scope

Brownfield: Integration of existing systems with agents

Greenfield: Design from agent-first perspective