AWS Launches CloudWatch Generative AI Observability Preview
AWS introduces CloudWatch generative AI observability to monitor and troubleshoot complex AI applications, providing end-to-end tracing across LLMs and agents.
Amazon Web Services (AWS) has unveiled a new generative AI observability feature for Amazon CloudWatch, designed to help organizations monitor and troubleshoot complex AI applications. The preview, announced in a blog post, addresses the challenges developers face when deploying large language models (LLMs) and AI agents.
Key Features
- End-to-end tracing: Provides visibility across LLMs, agents, knowledge bases, and tools.
- Flexible deployment: Works with Amazon Bedrock AgentCore, EKS, ECS, and on-premises infrastructure.
- OpenTelemetry support: Compatible with frameworks like Strands Agents, LangGraph, and CrewAI.
- Integration with CloudWatch: Leverages existing features like Application Signals, Alarms, and Dashboards.
Implementation Scenarios
The blog outlines two deployment scenarios:
- Agents hosted on Amazon Bedrock AgentCore runtime: Uses the AWS Distro for OpenTelemetry (ADOT) SDK for automatic instrumentation.
- Agents running outside AgentCore: Requires manual setup of OTEL environment variables.
Observability Dashboard
The CloudWatch console now includes a GenAI Observability (Preview) section with:
- Model Invocations dashboard: Tracks metrics like invocation count, token usage, and error rates.
- Bedrock AgentCore Agents dashboard: Monitors runtime metrics, sessions, and traces.
Benefits
- Centralized monitoring: View all AI agents in one place.
- Performance insights: Analyze latency, errors, and throttling.
- Troubleshooting: Drill down into individual traces and spans.
For more details, visit the official documentation.
Related News
AWS extends Bedrock AgentCore Gateway to unify MCP servers for AI agents
AWS announces expanded Amazon Bedrock AgentCore Gateway support for MCP servers, enabling centralized management of AI agent tools across organizations.
Agentic AI Transforms Asset Management in 2025
Asset management is undergoing a transformation with AI agents that streamline complex tasks under human oversight, shifting from pilots to integrated solutions.
About the Author
Alex Thompson
AI Technology Editor
Senior technology editor specializing in AI and machine learning content creation for 8 years. Former technical editor at AI Magazine, now provides technical documentation and content strategy services for multiple AI companies. Excels at transforming complex AI technical concepts into accessible content.