Skip to main content

AWS Bedrock

Overview​

AWS Bedrock is the default LLM provider for OpsWorker's SaaS deployment. OpsWorker uses a multi-model strategy through Bedrock to optimize investigation quality and speed.

How It's Used​

OpsWorker uses different AI models for different investigation stages:

StageModelPurpose
Routing, field extraction, summarizationAmazon NovaFast parsing, routing, and summarization of alert data
Deep analysisClaude (default us.anthropic.claude-3-5-sonnet-20241022-v2:0; also Claude Sonnet 4 / 4.5 and Haiku 4.5)Root cause analysis and recommendation generation
Chat responsesClaudeInteractive Q&A with context

This multi-model approach balances speed and depth: fast models handle routing and extraction, while more capable Claude models handle complex analysis. (Amazon Titan is not used.)

Configuration​

SaaS Deployment​

No configuration needed. OpsWorker's SaaS platform uses AWS Bedrock by default. All AI processing is handled automatically.

Dedicated Deployment​

For dedicated deployments in your own AWS account, you can use your own Bedrock access:

  • Models are deployed within the same AWS region as your OpsWorker instance
  • You control model access and usage through your AWS account
  • Contact the OpsWorker team for setup

Data Processing​

  • Investigation data is sent to AWS Bedrock for AI analysis
  • Processing occurs within AWS infrastructure
  • Data is not used for model training
  • Data retention follows OpsWorker's data handling policies

Next Steps​