Architecting the Future: Modern Solution Design with Agentic AI
summary
Claims that Agentic AI modernizes decades-old workloads by orchestrating multi-agent workflows on AWS (SageMaker, Bedrock, S3, OpenSearch, PGVector) using LangChain/LangGraph, Q Developer, Cursor and an MCP as smart API. Demos show automated code analysis and generation — 40,000 lines of COBOL analyzed across 159 file types and Java code plus unit tests produced quickly — and business flows where supervised agents cut loan processing from 30–60 days to ~1 week. Systems increase productivity but remain human-in-loop, not fully autonomous.
AI-written summary · about
chapters (15)
- 0:00Introduction and welcome
- 2:40Evolution of software architecture
- 6:15Traditional AI vs Agentic AI
- 9:07Agentic AI principles and memory
- 13:40Model Context Protocol (MCP) overview
- 16:31Multi-agent collaboration architectures
- 18:40Loan application processing demo
- 23:45Modernizing legacy COBOL applications
- 27:50Agent model selection and optimization
- 30:15Open source shift in AI models
- 32:03Q developer and code generation
- 36:22Code analysis and modernization tools
- 37:45Top use cases for Agentic AI
- 39:31Future of goal-driven AI systems
- 40:58Q&A and closing remarks
related talks
20mReimagining InnerSource for the Agentic AI Era
44mAgentic AI at Sage with AgentCore
42mAgentic AI and the Future of AI
29mAgentic AI: What It Is and Why It Matters
33mA View from the Top: Introducing Developers to the Future of Software with Agentic AI
27mSoftware engineering with GenAI
27mAgentic Engineering: Working With AI, Not Just Using It
27mMulti-Agent AI: Transforming Business Operations and Value Creation