How it works
Four phases. One continuous cycle.
01
Pre-assessment
The system asks one quick question. If you already know the topic, the explanation is skipped and you go straight to practice.
02
Explanation
The AI explains the topic through Feynman-style dialogue: asks clarifying questions, checks understanding. Readiness is shown on a progress bar.
03
Practice
10 exercise formats adapt to your level. Difficulty grows with your mastery score — from easy questions to complex applications.
04
Spaced repetition
The SM-2 algorithm computes exactly when to review each topic. You never repeat what you still remember well.
Features
Everything you need for deep learning
Semantic evaluation
AI evaluates the meaning of the answer, not keyword matching. Understands non-standard phrasing and partial answers.
Knowledge graph
Topics are linked in a dependency DAG. The next topic unlocks when the previous one is mastered ≥20%.
Bring your own key
Connect your own OpenAI, Anthropic, or any compatible provider API key. Credits are not deducted.
Interleaving
Tasks from different topics are interleaved. The brain must retrieve the right approach each time — this strengthens memory.
Progress analytics
Mastery score, progress curve, next review time — for each topic. A summary dashboard across all courses.
Courses in 30 seconds
Enter a topic — AI generates the course structure, topics and prerequisites. Edit as you like.