An Agent Walks Into a Bar... and Meets Another Agent: Multi-Agent Systems in Action
summary
Argues that scaling from one LLM-driven chatbot to coordinated teams avoids prompt bloat and enables specialization, illustrating four concrete orchestration patterns: sequential, concurrent, group chat, handoff. Recommends Microsoft Agent Framework (open source, .NET and Python only) plus Semantic Kernel and AutoGen heritage, cites experience across 100+ enterprise customers, and lists tradeoffs: parallelism and fault tolerance versus communication overhead, context-window limits, and security/permission complexity. Covers agent-to-agent protocol, long-running flows, testing each agent, observability, and least-privilege auth.
AI-written summary · about
chapters (9)
- 0:00Multi-agent systems intro
- 3:37Defining AI agents
- 10:05Bar analogy for orchestration
- 16:08Orchestration challenges
- 27:08Choosing orchestration patterns
- 33:06Tools for multi-agent systems
- 40:56Best practices for agents
- 42:22Future trends in agents
- 43:48Call to action
related talks
41mPatterns & Practices for building Multi-Agent Systems
28mThe Put That Sh*t On Agent: Building Multi-Agent AI Systems That Actually Talk to Each Other*
32mBuilding Multi-Agent Systems: From Concepts to Collaboration
50mScalable Innovation: Exploring Multi-Agent Systems
26mAI powered automation & multi-agent orchestration
27mMulti-Agent AI: Transforming Business Operations and Value Creation
56mBuilding AI Agents and Multi-Agent Systems with Azure AI
56mDesign Patterns for Multi Agent Systems