The Pedagogical Core (v12) Enabled

Education Strength as Identity.

Built on 6 cognitive learning-science principles: struggle before solution, active retrieval, confidence calibration, spaced mastery gating, interleaved practice, and misconception diagnosis.

Groq + Gemini Gateway35s Timeout
50 Unit Tests100% Passed
Supabase pgvectorHNSW Index
Pedagogical Engine Active

Topic Mastery

Quantum Mechanics

Streak🔥 5 Days
💡 Principle 1 (Struggle First): What mathematical operation isolates acceleration when mass is multiplied by acceleration?
🔒 Principle 4 (Mastery Gated): 85% score in session 1! 1 more review needed spaced 24h apart.
Cognitive Science Corev12 Standard

The 6 Pedagogical Principles

Designed around how human memory works, not just how AI generates text.

🧩

1. Struggle Before Solution

Fresh doubt threads default to Socratic scaffolding (hints/questions) rather than giving away direct answers immediately.

🧠

2. Retrieval Before Review

Forces active text recall attempts before revealing stored flashcard back answers to maximize long-term retention.

🎯

3. Confidence Calibration

Captures confidence ratings before quiz answers are revealed and flags overconfidence (Illusion of Competence).

🔒

4. Multi-Session Mastery Gating

Holds topics in progress until N correct reviews are verified across at least 2 separate sessions spaced 24h apart.

🔀

5. Interleaved Practice

Round-robin alternates study items across multiple active topics rather than drilling in blocked single-topic sessions.

6. Misconception-Aware Correction

Analyzes the specific wrong option chosen to explain the plausible misconception behind that specific answer.

System Architecture7 Core Engines

Enterprise System Architecture

Built on Next.js 16, Supabase pgvector, Groq, and Upstash Redis.

Persistent Learner Memory Engine

Maintains long-term student memory in Supabase tracking per-topic mastery scores, mistake patterns, and preferred explanation style.

RAG Material Grounding

Uses Supabase pgvector HNSW vector search over uploaded syllabi & notes with explicit section citations.

Verification & Audit Layer

Independent AI verification audit pass and multi-step arithmetic code execution in sandboxed environments.

Multi-Provider AI Gateway

Groq (70B/8B) + Gemini 2.0 Flash with automatic provider failover and 35-second provider timeouts.

SM-2 Spaced Roadmaps

Persistent 7-day study paths with SuperMemo-2 review schedules, streak tracking, and mastery badges.

Outcome Data Flywheel

Logs quiz outcome telemetry to continually track which teaching strategy delivers highest correctness rate per topic.