Observability in AI-Powered Applications with Amazon Bedrock
summary
Shows a live serverless demo: a Lambda Function URL that calls Amazon Bedrock models (Llama 3, Mistral, GPT) and sends telemetry into New Relic so traces show each hop, token usage and per-hop duration breakdown. Claims full AI observability with model inventory and per-model cost accounting, supports response streaming and OpenTelemetry export. Repository and deployment steps exist; the demo runs and surfaces errors, latencies and token counts so teams can spot hallucinations and costly models.
AI-written summary · about
chapters (14)
- 0:00AI observability introduction
- 3:02AI evolution and workflows
- 5:16AI model security
- 6:23AI monitoring challenges
- 8:35AI observability layers
- 9:20Bedrock serverless demo
- 10:23API invocation demo
- 11:26New Relic AI monitoring
- 12:34AI tracing and tokens
- 14:21Model performance analysis
- 16:07AI dashboards and costs
- 17:53Multiple AI providers
- 18:58Resources and raffle
- 19:59Q&A and OpenTelemetry
related talks
28mMeasuring Serverless Apps in Amazon Bedrock
29mElevating generative AI and LLM observability
55mSecurely build generative AI apps & control data with Amazon Bedrock
24mBuilding GenAI applications with Amazon Bedrock
39mGenerative AI Solution with AWS Serverless and Bedrock
31mLamRAG: From Data to Constructive Insights using Amazon Bedrock
41mServerless LLM with AWS Lambda and LangChain: A Revolution in Application Development
39mModel choice in Amazon Bedrock