Skip to main content

Release Notes - December 2025

Highlights​

December 2025 introduces ML Hub, a comprehensive in-database machine learning platform, bringing MLOps capabilities directly into your conversational analytics workflow.

New: ML Hub​

Train, experiment with, and monitor ML models entirely through natural language - no notebooks, no MLflow configs, no infrastructure headaches:

  • Conversational Model Training: Train XGBoost, LightGBM, and Random Forest models by describing what you want to predict
  • Model Registry: Full lifecycle management with staging, production, and archived states
  • Experiment Tracking: Compare algorithms and hyperparameters scientifically, promote winners to production
  • Feature Store: Centralized feature definitions with entity-based lookups and feature sets
  • Model Monitoring: Automatic drift detection, performance tracking, and two-tier alerting
  • Scheduled Predictions: Batch predictions on cron schedules with email delivery

Key Capabilities:

  • Classification and regression with industry-standard algorithms
  • Automatic versioning with one-click promotion
  • Real-time and batch predictions through conversation
  • Monitor performance, data drift, prediction drift, and data quality
  • Visual ML Hub UI for model management and experiment comparison

Example:

Train an XGBoost model to predict customer churn using the
customer_data table. Use age, tenure, and monthly_charges
as features and churned as the target.

Learn more about ML Hub

Improvements​

  • Forge: Enhanced DAG visualization with improved real-time execution monitoring
  • Performance: Optimized model inference latency for production predictions
  • UI/UX: Updated ML Hub interface with improved experiment comparison charts

Coming Soon​

  • AutoML for automatic algorithm selection
  • Deep learning model support (TensorFlow, PyTorch)
  • Feature lineage visualization
  • Model explainability dashboards

Published: December 29, 2025