H2O.ai expands with Driverless AI

Driverless AI was H2O.ai's bridge from an open source toolkit for specialists into a higher value enterprise product for teams that do not have enough data scientists. Instead of asking users to write Python or R and tune models by hand, H2O.ai packaged data prep, feature engineering, model selection, and deployment into a guided workflow that business analysts could use, which widened the buyer inside a bank or insurer from the central data science team to many operating teams across the company.

• The jump matters because H2O.ai started as infrastructure for technical users running distributed ML workloads. Driverless AI moved the company up the stack from helping experts build models faster to helping non experts get a usable model at all, which supports premium enterprise pricing and larger platform deals.
  1 sacra
• This product move put H2O.ai into more direct competition with DataRobot and Dataiku. DataRobot centered its product on browser based automated model building for analysts, while Dataiku built a GUI that bundles data ingest, prep, AutoML, and app creation for domain experts in banking, life sciences, and manufacturing.
  3 sacra 4 sacra 5 sacra
• The practical wedge was strongest in regulated industries. H2O.ai already had traction with customers like Capital One, Wells Fargo, and Kaiser Permanente, and about 40% of revenue came from financial services, so automating model building let those customers spread ML into fraud, risk, and operations teams without hiring a large bench of specialists.
  1 sacra

The next step is for automated model building to become one layer inside a broader enterprise AI control plane. As AutoML features become table stakes, the winning platforms will be the ones that combine easy model creation with governance, deployment, and secure on premises operation for large regulated organizations.