v3.12.0

This release introduces significant new capabilities for custom logic execution, dataset management, and enhanced security for agentic systems.

What’s New

  • Dataset Generators visible in UI: Users can now manage and see dataset generators directly in the platform under the Evaluate section.

    example visualisation of a custom data set in the Evaluate UI

Improvements

  • Rebranding: All “AI GO!” references have been rebranded to AI Platform across the UI and CLI.
  • Dataset Visibility: Generated datasets now clearly display the generator used in the title section.

Bug Fixes

  • Model Inference: Fixed a critical issue where model usage from Python code failed due to missing inference IDs.
  • User Management: Resolved a validation error that prevented disabling or enabling users via the UI.
  • Evidence Table: Fixed a bug where null bytes in LLM-generated evidence caused Postgres transactions to roll back, breaking the evidence table.
  • Cache Integrity: Fixed issue when reusing solver-output caches which previously left UI scores empty.