Farewell to Siloed Dashboards: Amazon CloudWatch Omni Brings Collaborative, AI-Powered Observability to Life

Hi everyone, it's Shii-chan! I've got some exciting AWS news to share with you today!

AWS Blog aws.amazon.com

What was announced?

AWS Blog announced Amazon CloudWatch Omni, a new observability service. The post I read is titled "collaborative AI-powered observability for your applications," and within the broader CloudWatch Omni story, this piece focuses specifically on how you monitor your applications as a whole.

CloudWatch Omni gives your whole team access through a dedicated URL, separate from the AWS Management Console. You sign in with enterprise SSO through IAM Identity Center (which supports providers like Okta and Azure AD), so engineers can join in even without AWS Console access.

Why it matters

Until now, observability tended to be scattered across separate dashboards for each service and each team, which made it hard to quickly grasp how an entire application actually connects together. On top of that, teams are now running AI agents alongside their applications, adding even more surface area to monitor and pulling visibility further apart.

CloudWatch Omni is built to tackle exactly that problem: observability for apps and AI agents living in separate silos, and different teams staring at different screens. The big idea here is getting everyone looking at the same map while they talk things through.

What changes

With CloudWatch Omni, service-to-service connections (the topology) are automatically discovered from telemetry data and AWS Config information, so you can see how the whole system fits together without drawing architecture diagrams by hand. When you deploy new services, that topology updates automatically too.

What I find especially nice is that you can just ask questions in plain language. CloudWatch Omni analyzes your telemetry and surfaces insights in response, which means people who don't know a special query language can still jump into an investigation.

And the AWS DevOps Agent joins right in on those investigation sessions, correlating signals and suggesting next steps. Having human engineers and an AI agent collaborating in the same investigation session is what I think makes this service genuinely interesting.

Dive Deep

CloudWatch Omni organizes things around a concept called "spaces," where a team can group together the applications they own. Since telemetry gets organized per space, teams with different areas of responsibility can share just the scope that's relevant to them.

You can also declare the things you actually care about, like availability targets, latency budgets, and error thresholds, and monitoring adjusts to follow those declared standards. Instead of manually curating dashboards, you set your standards and the system keeps up as your architecture changes.

Under the hood, it's built on OpenTelemetry, so if you're already sending telemetry to CloudWatch, it works without reconfiguration, and OTLP endpoints let you bring in data from non-AWS workloads too. It's also designed to coexist with your existing CloudWatch alarms, dashboards, and APIs.

Wrap-up

• Amazon CloudWatch Omni brings application and AI agent observability together into one collaborative view
• It automatically discovers service topology from telemetry and AWS Config data
• You can ask questions in natural language and get insights analyzed from your telemetry
• The AWS DevOps Agent joins investigation sessions to correlate signals and suggest next steps
• "Spaces" let teams organize their own scope, and a dedicated URL with SSO makes it easy for everyone to join in

If you're an engineer or SRE who wants your whole team chasing down "what's happening right now" together, across services and teams, this announcement is definitely worth a look!