Customizing genAI applications for your business
summary
Practical blueprint for making foundation models business-ready: use Amazon Bedrock’s managed retrieve-and-generate API or the retrieve API plus knowledge bases to run RAG (automated chunking, citations, session context) and cut hallucinations and latency; use agents to query databases for real‑time answers. Embeddings: Titan text embeddings (1,536‑dim vectors for up to 8,000 tokens) and Coherent (1,024 dims for 512 tokens). Supported stores: OpenSearch Serverless, Aurora, Redis, Pinecone (MongoDB soon). Fine‑tune with labeled examples or use continued pre‑training on large unlabeled corpora; data stays in customer VPC and is not shared.
AI-written summary · about
chapters (5)
- 0:00Introduction to customization
- 8:52Overview of Amazon Bedrock
- 10:57Retrieval augmented generation deep dive
- 23:23Fine-tuning and pre-training
- 28:19Summary and closing
related talks
48mHow to Make Smart Architecture Decisions when Building Gen AI Apps
24mBuilding GenAI applications with Amazon Bedrock
28mImproving responses in generative AI applications
19mMind your business: Secure your generative AI application on AWS
38mGenerative AI - Architectures and applications in depth
47mOutsmarting the Genie: Smart Architecture Decisions for GenAI Apps
23mScaling GenAI Applications and Agentic AI for Business Competitive Differentiation
20mBuilding a secure end-to-end generative AI application in the cloud