Evaluations have been significantly enhanced and now support individual ratings for each chip. In addition, chip-level interactions have been improved, targeted sentence regeneration has been introduced, chips within the same category are handled more intelligently, and manual text edits are preserved more reliably.

The AI-generated evaluation text has been refined to create more differentiated and context-appropriate assessments. Repetitive phrasing has been reduced, punctuation has been improved, attribution handling has been refined, and overall wording has been further polished.

Change tracking has been expanded. The history now records adjustments to chips and ratings, manual text edits, and changes to the structure of evaluation sections in greater detail.