Demystifying AI Agents
summary
Claims that AI agents extend LLMs via tool-calling and add memory, planning, reasoning; demonstrates Java examples using langraph4j state graph to model nodes, transitions and compile a deterministic workflow that prints the graph; argues real deployments must handle CAP theorem tradeoffs and distributed data, mentions MCP and A2A protocols, Autogen, Crew AI and LangChain4j, warns about hallucination, security and human‑in‑the‑loop, and tests whether the 2025 “everybody” prediction actually shows up in August.
AI-written summary · about
chapters (12)
- 0:00Introduction and agenda
- 2:27Speaker background
- 6:00Fundamentals of AI agents
- 18:54Multi-agentic architecture
- 22:31Terminology and definitions
- 23:31Defining agentic systems
- 24:34Java coding example
- 26:40Distributed computing concerns
- 30:14System design and CAP
- 30:34PACELC and GenAI
- 31:06Design patterns, challenges, and conclusions
- 45:22Conference closing
related talks
34mAI Agents: Transforming Autonomous Decision-Making
28mAI Agents Graph: Your Following Tool in Your Java AI Journey
50mIntelligent Agents, the New Frontier of LLMs
41mBringing stories to life with AI, data streaming and generative agents
29mBuilding Trust in AI Agents
02h 08mIntroduction to AI Agents
01h 03mAgents Everywhere
31mAgentic AI architecture patterns