AWS Launches CloudWatch Omni for Collaborative AI-Powered Observability

Summary

Amazon Web Services has launched CloudWatch Omni, a dedicated web experience for collaborative observability of applications and AI agents. Teams access it through an organization-specific URL with enterprise single sign-on via IAM Identity Center, including providers such as Okta and Microsoft Entra ID, without needing AWS Management Console access. Built on OpenTelemetry, Omni displays telemetry already sent to CloudWatch without reconfiguration and can receive telemetry from other instrumented environments through an OTLP endpoint and connectors. It organizes logs, metrics, traces, and alarms around applications, automatically discovers services, maps dependencies, and updates application topology as systems change. Teams define availability targets, latency budgets, and error-rate thresholds, while dynamic views reduce the need to maintain static dashboards and manually tune alarms. Each team can create a Space that groups its applications and points to existing CloudWatch data without additional data movement. During incidents, Omni opens an investigation session with topology and correlated signals, while Amazon DevOps Agent analyzes events, traces possible root-cause paths, preserves investigation history, and suggests mitigation steps. Engineers from different teams can join the same session with its existing context. The article’s example follows an elevated checkout error rate through a recent deployment, payment API latency, and a provider gateway configuration change before the team rolls back the cause. Omni also provides natural-language queries, service health alerts, and observability features for generative AI and agentic workloads, including trace exploration, evaluation frameworks, and real-time monitoring. Existing CloudWatch alarms, dashboards, APIs, and console workflows remain unchanged; Omni is now available, with pricing provided separately on the CloudWatch pricing page.