Multi-Agent Orchestration Patterns: Pick the Right Shape

Most multi-agent failures aren't model failures — they're shape failures. The wrong pattern for the problem. In this 37-minute workshop you'll learn the four canonical orchestration patterns — Parallel Dispatch, Sequential Pipeline, Supervisor plus Specialists, and Iterative Refinement — when each one fits, when each one is wrong, and how to diagram the right shape for your real problem. 🎯 What you'll be able to do by the end: ✓ Recognize the four canonical orchestration patterns wherever they appear ✓ Pick the right pattern using the three-axis selection matrix ✓ Compose hybrid patterns when one shape isn't enough ✓ Diagram nodes, edges, and contracts cleanly ✓ Write a one-page architecture spec for any system you design 📚 CHAPTERS 00:00 Cold open — shape failures, not model failures 04:52 Part 1 — Parallel Dispatch 09:59 Part 2 — Sequential Pipeline 12:45 Part 3 — Supervisor + Specialists 16:33 Part 4 — Iterative Refinement 23:33 Part 5 — Choosing the Right Pattern 29:16 Diagrams — conventions and worked example 33:26 Recap & the Module 4 assignment 35:45 Up next — Module 5: system design 🔧 LINKS & RESOURCES - Multi-Agent Architect course: https://[your-site].com - Module 3 (the prerequisite): https://youtu.be/[MODULE_3_ID] - One-page architecture spec template: https://[your-site].com/lessons.html - Claude Code documentation: https://docs.claude.com 🧠 WHO THIS IS FOR Engineers who shipped Module 3's mini-system and are ready to think in patterns. #ClaudeCode #AIAgents #SystemDesign #Anthropic #PromptEngineering #AgentOrchestration #MultiAgent #AIArchitecture