Practical Agentic RAG with SpringAI in Modern Enterprise Applications
summary
Practical recipe for an enterprise RAG + agent stack using Java 21 and Gemini Flash/Pro: SpringAI routes models, uses function calling for tools like weather and exchange rate APIs, and runs agents in Cloud Run. Documents are hierarchically or hypothetically chunked, embeddings stored in AlloyDB vector store, retrieval returns 1–10 chunks (demo uses five) with reranking and a 0.6 similarity cutoff, query compression and implicit caching cut costs, MCP and A2A enable tool and cross‑agent integration, and evaluation plus guardrails are required before production.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 3:03Workflow vs Agents
- 6:47How Agents Evolve
- 8:52Dynamic Agents
- 9:45Flows
- 10:55Function Calling
- 12:10Demo
- 15:00Retrieval
- 16:23RAG
- 20:51Testdriven AI development
- 27:13Legacy Data
- 27:43Model Context Protocol
- 35:56Agent Protocol
- 40:43Guard Rails
- 46:11Pitfalls
related talks
45mFrom RAG to Enterprise AI Agents: Building Intelligent Java Apps
49mBuilding AI Agents with Spring & MCP
26mJava’s GenAI Era: Getting Spring AI and Neo4j Applications to Production
01h 00mSpring AI
44mThe Spring AI Ecosystem in 2026: From Foundations to Agents
53mUnleash the Power of AI Agents with Java Building Intelligent Application
41mFrom Chat to RAG to MCP: Enhancing Java Applications with AI
56mRAG: from dumb implementation to serious results