AWS Launches CloudWatch Omni for Collaborative AI Observability

Overview

Amazon Web Services has officially launched Amazon CloudWatch Omni, a new AI-powered observability experience designed to help engineering teams monitor both traditional applications and complex AI agents within a unified workspace. The service represents a shift from siloed infrastructure monitoring toward an application-centric model that prioritizes collaborative investigation and automated context retention.

Access to Omni is provided through a dedicated URL specific to each organization, allowing engineers to sign in using existing enterprise identities via IAM Identity Center. This integration supports major identity providers such as Okta, Azure AD (Entra ID), and other SAML 2.0-compatible systems. Consequently, site reliability engineers (SREs), developers, and managers can collaborate in the same environment without requiring direct access to the broader AWS Management Console.

The platform is built on OpenTelemetry, ensuring that telemetry data already ingested by CloudWatch appears in Omni without reconfiguration. For workloads instrumented with OpenTelemetry outside the immediate AWS ecosystem, the service accepts data via the OpenTelemetry Protocol (OTLP) endpoint, creating a cohesive view of distributed systems.

Key Capabilities

CloudWatch Omni addresses three primary challenges cited by engineering teams: fragmented tooling, manual dashboard maintenance, and loss of contextual continuity during incident escalation. The service aims to consolidate these pain points into a single, adaptive interface.

• Collaborative Investigation Environment: By centralizing access through enterprise SSO, Omni ensures that every team member—whether a database engineer or a product manager—views the same data and investigation context. When an issue escalates, the next responder joins an active session with the full history of findings already visible, eliminating the need to reconstruct events from scattered Slack threads or screenshots.
• Adaptive Application Topology: The system automatically discovers services and maps dependencies as applications evolve. Instead of manually curating static dashboards or tuning individual thresholds, teams declare their Service Level Objectives (SLOs), availability targets, and error rate budgets. Omni then adjusts its monitoring posture automatically when new services are deployed or infrastructure changes occur.
• AI-Powered Root Cause Analysis: Integrated with Amazon DevOps Agent, the platform participates in investigation sessions alongside human engineers. The agent correlates signals across services, traces root cause paths through the dependency graph, and suggests actionable next steps. Because the agent operates from the same telemetry sources as the engineers, its recommendations are grounded in the real-time state of the application.

The Investigation Workflow

A typical incident workflow in CloudWatch Omni demonstrates how the platform reduces the time between detection and resolution. When an alarm triggers—such as elevated error rates in a checkout service—Omni automatically opens an investigation session pre-loaded with relevant context.

1. Initial Correlation: The session displays the service topology alongside correlated signals, such as a deployment that occurred ten minutes prior or increased latency from a downstream payment API. Amazon DevOps Agent provides an initial analysis of these connections.
2. Deep Dive: The on-call SRE confirms the deployment correlation and pulls up trace views to identify specific failing endpoints. They may also check whether the payment API latency correlates with a known capacity limit.
3. Collaborative Resolution: If the issue requires specialized knowledge, the SRE escalates to the payments team. A payments engineer joins the same session, seeing all prior findings. They might identify that a configuration change in the payment provider’s API gateway is the root cause and proceed to roll back the change.
4. Automated Documentation: The entire investigation history, including decisions made and actions taken, is captured automatically. This creates a comprehensive record for post-incident review without the need for separate manual reporting.

Getting Started and Setup

Organizations can begin using CloudWatch Omni directly from the CloudWatch console by selecting the option to try the new feature. The setup process is designed to minimize friction and integrate seamlessly with existing identity management infrastructure.

Identity Configuration

The first step involves connecting your identity provider through IAM Identity Center. This step is critical for enabling secure, passwordless access for team members. Once connected, your team accesses Omni via their dedicated organization URL, ensuring that governance policies are enforced without exposing underlying AWS Console permissions.

Creating a Space

After identity setup, users create a Space, which serves as the central workspace for a team. A Space groups the applications owned by that team along with their associated telemetry. This logical grouping allows different departments to maintain separate observability contexts while sharing the same underlying infrastructure.

• Automatic Discovery: Upon creation, Omni scans your telemetry data and AWS Config resources to automatically discover services. It then maps the dependencies between them, rendering an immediate visual representation of your application topology.
• Natural Language Interaction: Teams can query their telemetry using plain English questions. Omni analyzes the data and surfaces insights, reducing the need for complex query languages or pre-built report generation.
• Alert Configuration: Users can define service health alerts and specify what metrics matter most to their team. If an anomaly is detected, the system can trigger an Amazon DevOps Agent investigation to diagnose the root cause and propose mitigation plans.

Application-Centric Organization

A defining feature of CloudWatch Omni is its departure from infrastructure-centric monitoring. Rather than organizing data by individual servers or databases, the service structures observability around applications. This approach reflects how modern engineering teams actually work and fail.

Each Space points to existing CloudWatch data—including logs, metrics, traces, and alarms—without requiring additional data movement or migration. This compatibility ensures that historical data remains accessible while providing dynamic views that replace the maintenance burden of static dashboards. The result is ongoing visibility into SLO compliance and application health that adapts as the system changes, allowing teams to focus on resolution rather than configuration.