Adaptive Socratic AI Tutor

Design "[INSERT PRODUCT NAME]", an adaptive AI tutor for [INSERT SUBJECT AND AGE RANGE, e.g., algebra for ages 12 to 14]. Present the pedagogical design first; build modules only after my approval. 1. LEARNING MODEL: mastery-based progression with a prerequisite graph; define competencies and evidence thresholds. 2. TUTOR PERSONA: Socratic, patient, growth-minded; adapts language to age; never simply hands over answers. 3. DIAGNOSIS: initial placement, misconception detection from error patterns, and a persistent per-student skill matrix. 4. ADAPTATION: adjust difficulty, hint scaffolding in layers, and representations across visual, symbolic, and verbal modes. 5. CONTENT: an item bank tagged to [INSERT CURRICULUM STANDARD], worked examples, and spaced retrieval practice. 6. INTERACTION: step-by-step problem solving, handwriting and math input, voice for young readers, encouragement without praise inflation. 7. DATA FLOW: attempts to an evidence tracker to a next-best-action selector; teacher and parent dashboards with role-based views. 8. GUARDRAILS: refuse to complete graded assessments outright, watermark AI assistance, and detect learned helplessness. 9. EQUITY: multilingual support, low-bandwidth mode, and bias review of word problems for culture, gender, and names. 10. SAFETY: COPPA-aligned chat filtering, no open-ended social chat, and human-escalation flags. 11. PRIVACY: school data agreements, minimal collection, no ad targeting, retention tied to enrollment, delete-on-exit. 12. EFFICACY: pre/post tests, time-to-mastery, correlation with grades, and A/B experiments with pedagogic logging.