Back to blog home

Fiddler Integration for Datadog: Monitor ML Metrics That Matter in Datadog

When it comes to building and deploying ML models, accuracy and trust is just as important as performance. With more and more models embedded into business-critical applications every day — especially with all the recent AI breakthroughs — AI-forward companies need a way to observe the health of their ML systems the same way they do their business applications. 

IT organizations rely on application performance management (APM) platforms to be the centralized command center for all their application monitoring, and, with the proliferation of ML models, they are increasingly incorporating the MLOps lifecycle into their existing workflows. 

To help these cutting-edge companies on their responsible AI journeys we are excited to announce our integration with Datadog, a leader in APM! We are providing Datadog customers with powerful ML insights generated by the Fiddler AI Observability (formerly Model Performance Management) platform, right within their Datadog console.

IT organizations now have telemetry on both application and model performance for a comprehensive view of IT performance, helping them troubleshoot issues quickly from a central dashboard.

Visualize model performance metrics generated by Fiddler in a Datadog dashboard
Figure 1: Visualize model performance metrics generated by Fiddler in a Datadog dashboard

 As the AI Observability pioneer, Fiddler gives companies visibility into the models that power their AI applications.

How the integration works

The Fiddler-Datadog integration enables IT and MLOps teams to push model metrics calculated by Fiddler into their Datadog console, helping them save time monitoring relevant performance metrics and ensuring the health of their ML systems. Datadog users can filter down to the projects, models, or metrics that they care most about when investigating model performance issues. Once an issue is surfaced in Datadog, ML teams can use Fiddler’s best-of-breed model monitoring and explainable AI to drill-down and perform root cause analysis to resolve the issue quickly.

Figure 2:  Monitor model performance metrics covering performance, drift, data integrity and traffic in Datadog

The integration installs into the Datadog Agent and moves model metrics from the customer’s Fiddler environment to their Datadog environment at a configurable cadence.

Haven’t used Fiddler yet? Request a demo