Create evaluation records from LLM scores and narratives

Your interpreter submissions send raw AI scores and notes, making tracking and feedback inconsistent. Structured score and narrative records are created automatically so reviewers can act same day.