AWS Launches CloudWatch Omni for Unified AI Agent Observability

AWS says traditional observability tools, including CloudWatch, will struggle to explain why AI agents and agentic applications behave as they do in production. [1] CloudWatch uses metrics, logs, and traces to monitor applications and infrastructure across AWS accounts, regions, and services through the AWS Management Console. [1]

According to AWS, understanding agent behavior currently requires teams to move among agent-specific tools in Amazon Bedrock AgentCore, application performance monitoring, and CloudWatch infrastructure monitoring. [1]

CloudWatch Omni is a new off-console experience that combines agent, application, and infrastructure telemetry in an application-centric setup for investigating agent behavior in context. [1] AWS says developers and operations teams can begin with the application under investigation instead of navigating individual AWS resources and monitoring consoles. [1]

The tool automatically discovers application topology and component connections, allowing teams to query telemetry immediately with natural language or SQL and without manual setup. [1]

CloudWatch Omni changes the starting point for an investigation from individual resources to the application and its connected components. Rather than treating CloudWatch infrastructure views and Amazon Bedrock AgentCore’s agent-specific tools as separate investigative stops, Omni is designed to place their relevant telemetry in one application-oriented experience. Natural-language and SQL access could also give teams different ways to examine the same automatically discovered topology. What the sources don't address: how accurately Omni explains complex agent decisions or how it will be priced.