Building AI Agents with Spring & MCP
summary
Shows a runnable Java stack that wires Spring AI to Amazon Bedrock models and a Postgres + PGVector store to build a RAG chat assistant that finds adoptable dogs, turns 18 embeddings into vectors (initial run ~4s, subsequent startup ~0.2s), maps LLM output to structured IDs, and executes actions via MCP tool calling over HTTP to schedule pickups. Demonstrates Amazon Bedrock (Nova + Cohere) as chat and embedding models, uses PGVector RAG embeddings, adds chat memory, and secures agent-to-tool calls with OAuth2; demos work after minor live fixes.
AI-written summary · about
chapters (13)
- 0:00Introduction to Spring AI
- 1:32Building the dog adoption demo
- 9:37Setting up the database and model
- 13:10Adding chat memory to the agent
- 14:32Configuring system prompts
- 15:56Implementing RAG for dog matching
- 20:30Scheduling dog adoptions with tools
- 23:41Introducing MCP protocol
- 29:03Securing AI agent communications
- 40:17Exploring MCP protocol benefits
- 44:10Creating authenticated MCP book server
- 48:04Performance benefits of Java in AI
- 48:45Q&A and closing remarks
related talks
40mBuilding AI Agents with Spring & MCP
27mBuilding AI Agents with Spring & MCP
53mHands On Building Agents
42mBootiful Spring AI
01h 00mSpring AI
54mFrom Single-Shot LLMs to Intelligent Agents: Building Scalable AI Systems with Spring AI and MCP
50mModular RAG Architectures with Java and Spring AI
45mFrom RAG to Enterprise AI Agents: Building Intelligent Java Apps