Build AI applications with Guardrails for Amazon Bedrock
summary
Shows how Guardrails for Amazon Bedrock enforces four policy types and how to deploy them end to end: creates a banking assistant guardrail with deny topics and content and word filters, tests it live against Foundation Models (demo uses Anthropic Claude), blocks prompts for investment advice and toxic language, applies PII masking to redact names, and ships logs to CloudWatch. Also builds an agent that calls OpenAPI‑defined APIs via Lambda, applies the same guardrail, and demonstrates the guardrail successfully intercepting and steering responses.
AI-written summary · about
chapters (9)
- 0:00Introduction to session
- 1:26Challenges in generative AI apps
- 6:06What are Guardrails for Bedrock
- 11:56Live demo creating Guardrail
- 24:07Summary of Guardrails demo
- 25:12Introduction to Agents for Bedrock
- 31:36Demo creating an Agent
- 37:17Testing Agent with Guardrails
- 39:55Recap and next steps
related talks
18mBuild responsible AI applications with Guardrails for Amazon Bedrock
20mResponsible AI with Amazon Bedrock Guardrails
39mBuilding Safe AI Agents
31mUsing Guardrails.ai for building Generative AI apps
19mMind your business: Secure your generative AI application on AWS
17mSafeguarding sensitive data used in generative AI with RAG
38mBuilding AI responsibly with a GRC strategy
15mAI Guardrails for Enterprise Deployments