Skip to content
Stand with Ukraine flag

Trendz Analytics v1.13.x Release Notes

← Back to all releases

Improvements:

  • Redesigned anomaly autodiscovery tasks.
  • Added job for saving anomaly scores as telemetry in ThingsBoard.
  • Added ability to create alerts based on discovered anomalies.
  • Added new filter conditions — “not in” and “does not contain”.
  • Filter business entities based on user permissions.
  • Improved AI Assistant conversation interface.

Bug fixes:

  • Fixed fill gaps strategy during anomaly detection.
  • Fixed issue with failed topology rediscovery.
  • Fixed query planner for calculated fields.
  • Invalidate JWT tokens based on user activity.
  • Fixed multitenant validation procedure.

Improvements:

  • Added prompt templates for agentic knowledge and instructions management.
  • Added summarization and explanation for visualizations with AI Assistant.
  • Added conversation interfaces for AI Assistant.
  • Added ThingsBoard widget action to interact with AI Assistant.
  • Added support for OpenAI API-compatible models.
  • Added support for custom and self-hosted LLM providers.
  • Added support for OpenAI o4 family models.
  • Added Trendz task sequencing API.

Bug fixes:

  • Fixed heatmap translation.
  • Fixed AI Assistant memory aggregation.
  • Fixed drag and drop after unsuccessful view config save.
  • Fixed issue with renamed calculated fields.
  • Fixed manual task execution failures.

Improvements:

  • Added AI Assistant for creating visualizations.
  • Added AI Assistant widget for ThingsBoard dashboards.
  • Added configurable LLM providers for the AI Assistant.

Bug fixes:

  • Fixed access denied error for public dashboards.
  • Improved fill gap strategy for time series fields.
  • Fixed translations for chart tooltips.
  • Fixed task automated refresh for calculated fields.
  • Fixed delta aggregation for raw data loading mode.
  • Fixed issues with UNIQUE + COUNT operations in calculated fields.
  • Fixed issues related to model retraining in Prophet and multi-Prophet scenarios.
  • Corrected the AUTO segmentation strategy for prediction models.