Building Agentic AI applications on AWS
summary
Three AWS paths for agentic apps: out‑of‑box Amazon Q, fully managed Amazon Bedrock agents, or DIY with LangChain/SageMaker. Demo builds a trip planner with a Claude 3.5 orchestrator delegating to parallel subagents (restaurant scout, activity finder) that call Lambda web searches, DynamoDB and OpenSearch, then compile an itinerary; Bedrock Flows shows per‑agent latency and token usage. A second demo uses MCP (Model Context Protocol) with local Docker MCP servers (AWS Cost Explorer, Perplexity), Python + pandas auto‑generated code to produce a cost bar chart.
AI-written summary · about
chapters (7)
- 0:00Introduction and welcome
- 1:08Agentic AI overview and modes
- 2:34AWS options for agentic AI
- 6:07Amazon Bedrock agents deep dive
- 8:35Multi-agent collaboration demo
- 16:28Inline agents and MCP demo
- 27:32Key takeaways and resources
related talks
23mBuilding Agentic AI Systems with AWS Serverless
15mServerless & Agentic AI: Better Together
17mTransforming Robotics Service with Agentic AI on
19mBuilding Agents at Cloud Scale
23mBuilding Production Multi-Agent RAG Systems on Serverless AWS
16mBuilding Agentic AI applications with Positron and AWS Strands Agents
30mAccelerate AI to prod
43mThe AgentVerse - Build fast, Scale smarter