Agentic AI Patterns
summary
Distills concrete, reusable patterns for coordinating multiple models: four basic workflows — sequence, loop, parallel, routing — plus an autonomous supervisor that chooses and invokes agents at runtime. Demonstrates implementation with Quarkus and LangChain4j, structured outputs, function-calling tools, RAG, guard rails and shared agentic scope for context. Shows a routing demo where legal advice gets injected context from a medical agent and a supervisor demo that calls exchange/withdraw/credit agents and returns a computed recap, both working end-to-end.
AI-written summary · about
chapters (13)
- 0:00Agentic AI patterns intro
- 1:59AI services foundation
- 9:20Agentic vs single AI
- 10:22Rigid workflow patterns
- 15:59Loop workflows for agents
- 19:09Parallel execution workflows
- 20:13Conditional routing patterns
- 21:18Demo: routing & scope
- 31:34Supervisor pattern
- 35:25Demo: supervisor & currency
- 39:21Custom agentic patterns
- 45:02LangChain4j features & future
- 46:47Q&A
related talks
47mAgentic AI Patterns
51mAgentic AI Patterns
35mAI Agent Control Logic with Pattern Matching
27mEmerging Patterns Agentic AI
31mAgentic AI architecture patterns
31mBuilding and Orchestrating Real Production-ready Agentic AI Systems
46mAgentic Design Patterns Where You Are the Hero
24mAgentic Al in SW Development: Evolving Patterns & Protocols