StratusOn Crescendo Analytics: Agentic AI in action
Moka's Deployment Health dashboard, shown below, is powered by the StratusOn Crescendo Analytics service—a mesh of AI agents (called actors) deployed across all Azure regions, collecting real-time reliability, capacity and health analytics data from each Azure region.
This post presents a high-level overview of the service without digging too deep into its architecture or detailed design.
Moka's Deployment Health dashboard
What is StratusOn Crescendo Analytics?
StratusOn Crescendo Analytics (SCA) is an AI service built and maintained by StratusOn.
SCA is an Agentic AI system with autonomous AI agents performing specialized operations backed by our domain-adapted, fine-tuned AI models (custom LLMs). Architecturally, it follows what is known as the Actor Model, where it consists of a mesh of AI agents (called actors) deployed across all Azure regions. Each actor is specialized in a specific category of telemetry. Some of the data it collects is related to reliability, capacity and Azure service health, in addition to other data points. These are called static AI agents.
Fractal Forecast Service
The data is collected from the agents (actors), through a messaging protocol, and stored and processed by an AI service called Fractal Forecast (FFS) that runs predictive analysis and then models the region's deployment hygiene. This allows the SCA service to provide accurate data to Moka when contacted for the purpose of unit testing a deployment template in an Azure region.
SCA's Scalability
The service is massively scalable and capable of returning real-time analysis specific to the deployment template being analyzed within milliseconds for most Azure regions. It does that by spinning dynamic AI agents (actors) that are created on the fly and process the provided deployment template before creating other actors, as part of a workflow run inside our Maestro Orchestra workflow engine (currently used in our products Maestro Studio ENSEMBLE and the beta release of Maestro Studio SOLO), to perform sub-tasks and finally build a health report for the deployment. The health report is then displayed on the Deployment Dashboard's map (shown in both screenshots in this post).
Health Assessment at a Glance
The screenshot below shows what the health dashboard looks like when there are no potential issues related to the deployment template at the time the analysis report was generated. Clicking on any node on the map displays information about the node's health as it related to the current deployment. Moka's users can use the refresh button to refresh the data from SCA.
Deployment Health dashboard showing a healthy deployment template
Why SCA?
Anyone who has deployed on Azure using any Infrastructure-as-Code (IaC) framework will be able to relate to how frustrating it can be when they have a fully valid deployment template that has been tested many times only to fail on this one time, usually due to some capacity issue. Having zero-click access to StratusOn Crescendo Analytics from within Moka's UI enables getting an updated view of the deployment health posture—within seconds—before attempting to deploy, saving a user plenty of time and the hassle of having to clean up partial deployments after a failure.
This is the peace of mind many DevOps engineers have told us they wanted. Anyone who has productivity and deployment reliability in Azure as their top concern should care.
This is just a glimpse...
This is just a brief overview of the service. Keep an eye on our blog for a future blog post that dives deeper into the overall architecture which is largely based on the Actor Model.
In the meantime, if you have not done so already, subscribe to Maestro Studio AI and start interacting with Moka.
Get started at https://stratuson.ai.