Skip to main content
October 6, 2025
AI-powered protocol errors, design DOE refactor, and design optimization

AI-powered error handling for protocols

Protocol generation now uses AI to interpret and surface errors, making it easier to diagnose and fix issues without digging through raw logs. The overall protocol creation UX has also been streamlined.

Design parameter DOE refactor

The design parameter screen has been reworked to deliver a more classical design-of-experiments experience, making it easier to set up and manage parameter sweeps.Find out more →

Design optimization framework

The pipeline now includes a full design optimization framework with validation, error handling, and penalty mechanisms. Three example notebooks — fast charging, thickness-constrained capacity, and discharge capacity — are available to get started quickly.Find out more →
Improvements
  • Updated SQL experiment templates with improved error handling.
  • Switched parameterizations routing from v1 to v2.
Fixes
  • Fixed local database reset — restored missing permissions, roles, and templates data.
Improvements
  • Added design optimization parser for pipeline configurations.
  • Added cross-validated regression methods.
  • Added parse_experiment() for converting experiment configs into PyBaMM Experiment objects with parameter kwargs.
  • Added scalar_output and array_output properties to the Interactive optimizer.
  • Added phase option to InitialSOCHalfCell.
  • Added support for the “lumped” electrode type in interpolant calculations.
  • Improved initial stoichiometry calculations to use T_init.
  • Removed parameter scaling from CostLogger for cleaner cost tracking.
  • Simplified parameter normalization by removing normalized parameters.
Fixes
  • Fixed entropic change export bug in OCP interpolant.