Intelligent Agents, the New Frontier of LLMs
summary
Defines what makes a system an “agent” versus a plain LLM, outlines LLM limitations, and demonstrates building concrete agents in Java using LangChain4j and the Agent Development Kit (ADK). Shows agent design patterns that go beyond single LLM calls to fetch documents (RAG), trigger actions in surrounding systems, and discusses MCP (Model Context Protocol) and A2A (Agent-to-Agent) protocols.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:15IA agents with LLMs
- 5:11IA agent equation components
- 7:17IA agent planning workflows
- 15:00Function calling and environment
- 18:31Multi-agent system design
- 20:37Demo: Advanced RAG task splitting
- 34:44Demo: Sci-fi story generator
- 40:42Demo: ADK orchestration sub-agents
- 49:13Conclusion
related talks
01h 43mBuilding LLM Agents (Hackathon)
46mDemystifying AI Agents
36mLLMs and AI agents demystified
45mA comprehensive showcase of AI Agents and Agentic LLMs
41mBuilding LLM Agents – Challenges and Learnings
50mAutonomy of LLM Agents: Current State and Prospects
47mJava Meets AI: Integrating LLMs into Your Applications with LangChain4j
41mBringing stories to life with AI, data streaming and generative agents