Explore what’s possible with new AWS generative AI services
summary
Atos demonstrates a production AI contract insights pipeline that ingests Word/PDF/OCR text into S3/Postgres/OpenSearch, uses embeddings and a vector store with a RAG framework calling models via Amazon Bedrock (Claude 2, Titan) and SageMaker JumpStart, and exposes a UX that cites source pages; deployment moves from feasibility to MVP to production, cuts manual contract search across hundreds of thousands of agreements, and uses HELM for benchmarking while evaluating agents like LangChain/LlamaIndex and a Landing Zone Accelerator for secure governance.
AI-written summary · about
chapters (11)
- 0:00Introduction and speaker intro
- 1:05What is generative AI
- 2:30AWS generative AI portfolio
- 5:48Real-world use case introduction
- 7:38Technical architecture overview
- 10:30Behind the scenes RAG framework
- 12:17Landing Zone Accelerator benefits
- 14:25Key lessons learned
- 16:56Security and cost considerations
- 18:03Bedrock advantages summary
- 19:27Q&A and closing remarks
related talks
38mGenerative AI - Architectures and applications in depth
30mGenerative AI in action: from POC to business value
20mBuilding a secure end-to-end generative AI application in the cloud
24mBuilding GenAI applications with Amazon Bedrock
41mKeynote
35mReinventing with generative AI on AWS
01h 00mSecurely accelerating generative AI innovation
27mBuilding a Future-Ready AI Platform: Pictet AM Journey with AWS