AI 驱动 教育 运营

Example: 考勤, timetable, and assessment jobs are segmented into 确定性规则 (roster conflicts, room capacity, grade thresholds) and model-needed cases (plagiarism signals, nuanced performance risk, parent query intent). Most work routes to parsers and existing LMS/SIS logic. 高-volume patterns (common grade clusters, recurring timetable blocks) are distilled into small specialist models. Final execution — alerts, reports, compliance exports, and audit logs — runs as detached 系统 on campus infrastructure with zero token cost at runtime.