# Track 12: Agent Orchestration & Multi-Agent Systems

When one agent isn't enough: the topologies (orchestrator-worker,
supervisor, hierarchical teams, swarm/handoff), how agents actually
communicate and share state, and how the major frameworks differ. Ends
honestly — multi-agent adds real cost and failure surface, and a large
part of this track is knowing when *not* to reach for it.

## Modules

1. When One Agent Isn't Enough
2. Multi-Agent Topologies Overview
3. The Orchestrator-Worker Pattern
4. The Supervisor Pattern
5. Hierarchical Agent Teams
6. Swarm and Handoff Patterns
7. Sequential and Parallel Pipelines
8. Inter-Agent Communication
9. Shared State and Blackboards
10. Conflict Resolution and Consensus
11. Multi-Agent With LangGraph
12. CrewAI
13. AutoGen and Semantic Kernel
14. The OpenAI Agents SDK
15. Choosing an Orchestration Approach
16. Multi-Agent Failure Modes
17. Capstone Project