DataRobot, the leader in enterprise AI, today announced enhancements to its AI platform designed to further transform data into real business value. DataRobot’s latest release provides new capabilities for every step in the data science lifecycle, including next-level feature discovery, a new comprehensive autopilot mode to maximise accuracy, anomaly assessment insights, governed approval workflows for MLOps, and much more.

Today, most organisations recognise the opportunity that AI can bring. According to a recent research report conducted by MIT Sloan Management Review and BCG, nine out of 10 respondents agree that AI represents a business opportunity for their company, yet seven out of 10 companies reported minimal or no impact from AI so far. As organisations continue to adopt AI and scale the technology across the enterprise, they need an enterprise-grade platform that will help them move from “experimental AI”–or AI that isn’t value-oriented, readily operationalised, or trustworthy–and translate data to value across all of the AI lifecycle, from creation to consumption.

“For organisations to move beyond the ‘experimental AI’ phase, they need an AI partner that can help them deliver real business value,” said Phil Gurbacki, SVP of Product and Customer Experience, DataRobot. “As the only end-to-end platform on the market, we have the expertise and know-how to support our customers in driving real business change with their AI investments. Our latest enhancements further accelerate our ability to transform data into value.”

In the latest release of the platform, DataRobot’s headline capabilities include:

  • Next-Level Feature Discovery:An enhanced workflow to make the process of declaring relationships in data much easier. This improves how DataRobot will automatically discover valuable new features in primary and secondary datasets. In addition, users can now access logs for details on which features were explored, discarded, and generated. They can also now download the full training dataset, including the newly derived features.
  • Comprehensive Autopilot: The new comprehensive mode runs every single model in the repository taking as long as necessary to maximise accuracy when customers need it most. The new ‘Get More Accuracy’ feature allows them to start autopilot in an earlier mode such as ‘quick’ to get a baseline, then run comprehensive mode after they’ve seen initial results.
  • Anomaly Assessment Insights: This new interactive visualisation allows Automated Time Series users to quickly investigate anomalies and anomalous regions in their data, and see SHAP scores for the underlying features causing the anomaly. They can also explore anomalies through different time segments. By understanding the root cause of anomalies, users get actionable insights and enhanced explainability to justify their decisions.
  • Model Comparison, Reimagined: Based on best practices from DataRobot’s most experienced data scientists, the latest version of the platform has reimagined the way users compare models. It is now much easier and more intuitive to compare model, enabling users to be more confident that they have chosen the best one for deployment. DataRobot has enhanced its Lift Chart, ROC Curve, and Profit Curve, added support for more bins, and added new tooltips to enhance overall ease of use.
  • Governed Approval Workflows: Customisable governance policies and approval workflows enforce accountability for production-grade AI. This much requested new feature allows users to continue to deploy and manage production models while at the same time increasing the overall level of AI governance across their entire organisation.
  • Connect to Remote Repositories: Data science teams often govern and manage their models and model artifacts in popular open source code repositories and commercially available cloud services. DataRobot’s latest release allows users to connect directly to their GitHub and S3 repositories and dynamically pull model code and artifacts into DataRobot, making it simple to package, test, and deploy models in their production environment of choice.