The study OSwhere your videos live.

A student picks a target exam and a date. Flow builds the plan backwards from there — the AI ranks every topic in the syllabus, and weak areas get roughly 50% more time.

Every topic becomes two cards: a Learning card and an Assessment card. Cards move Mission → Battle → Conquered — and a failed check flips one back to Replay, with your video on it.

The AI tutor is context-aware — it knows the student's board, grade and the topic on the table. When a concept is visual, or a student keeps circling, it suggests a video.

Suggestions hydrate only from the approved catalog — the tutor can never fabricate a link. And every card lands at your clarity moment: mid-explanation, not mid-intro.

Drills run adaptive MCQs in batches of 10. Accuracy at 78% or above steps the difficulty up; 45% or below steps it down. Topics a student keeps missing get weighted heavily in the next batch.

When the session ends, up to 2 rescue videos attach to each weak topic. If your lesson is the clearest on that topic, it's the one that shows up.

For any chapter, the AI draws a concept map — 10 to 20 nodes, up to 3 levels deep. Nodes are typed — formula, example, common mistake — so the map teaches structure, not just vocabulary.

Your video pins to the exact node it explains. And with maps generated in 14 languages, a Tamil lesson can sit on a Tamil map.

Watch-through, completions at 90% or more, 1–5 ratings and helpful votes flow back into your catalog score — and into your dashboard.

The seven weights are public, and they decide tomorrow's recommendations. Your reliability grows with every approved lesson — great teaching compounds.