Most "AI agents" are a single LLM wrapped in a system prompt. They look great in demos and collapse the moment real users show up. I know because I built one for my 5,000+ member developer community, and the first version broke in ways I didn't see coming. In this talk, I'll walk through the multi-agent system I run in production: a tree of 12 specialist agents built on Google's ADK, coordinating through sequential pipelines, parallel fan-out, and LLM-driven dynamic routing, each armed with MCP tools to act on the real world. We'll cover the orchestration primitives, then climb into the real architecture: how onboarding chains three agents in sequence, how external knowledge fans out across GitHub, Dev.to, and StackOverflow in parallel, and how the root agent delegates per message. Then the fun part: the bugs. The drain-loop that cancelled a ParallelAgent mid-flight. The recency drift that surfaced 2020 articles in 2026. The callback layer I had to build for PII sanitization, caching, and observability ; none of which made it into the tutorial. You'll leave with a mental model for each pattern, production patterns that kept it running, and war stories that'll save you a week.
Muhammad Ahsan is a Google Developers Expert (GDE) in AI & Angular, an author of 4 world-wide published books, an Educator, and a Software Architect. https://bio.link/codewithahsan
