The platform helps you focus on solving problems with data science

Monitor with AI management platform


It’s expensive to maintain models in production: IT is monitoring the infrastructure, but no one is monitoring the behavior of the models.


Drift analysis

Quickly understand how real-world data changes over time.

  • Why? Data will change and that makes the initial model less and less reliable over time. The first step is to measure that change and then understand it to take proper action.
  • How? automatically computes a drift score, weighted by the impact of each feature in the model, to give a global evaluation of a model’s fit with its environment over time.

Champion / Challenger comparison

Safely update your models by comparing the behavior of your candidates beforehand.

  • Why? Even with the best monitoring infrastructure available, putting a new model into production is still a shot in the dark: you never know if some unexpected change in the data was not taken into account.
  • How? allows you to put two models side by side – one in production, the other silently being evaluated, so that you can estimate if the new model performs better than the previous one without interrupting production.

Usage & resources monitoring

Control your costs and understand how your users consume the model.

  • Why? Machine learning is expensive—even a simple model requires more resources than a traditional application. Monitoring consumption of resources and associated costs is a necessity in order to meet ROI requirements.
  • How? provides detailed information about your model’s resource utilization to understand peaks and valleys in order to best optimize your environment.