Why this method works

Computer-supported homework with scaffolding produced substantially better learning than traditional paper homework (effect size 0.61).

Here: Every question gives instant feedback, a hint ladder, and a final worked example only when needed.

Mendicino, Razzaq & Heffernan (2009), ASSISTments study · Source

Meta-analyses of intelligent tutoring systems show that worked examples and step-based feedback are among the strongest predictors of success.

Here: Each exercise includes graded hints ending in a full worked solution students can study after trying.

ITS K-12 meta-analysis (2024–2025) · Source

A carefully designed AI tutor produced more than double the learning gains of an excellent active-learning classroom in a Harvard RCT.

Here: Our tutor guides with questions instead of giving away full answers, even when students ask directly.

Kestin et al. (2025), Scientific Reports · Source

Spacing maths practice across sessions improves test performance and reduces overconfidence compared with massed practice.

Here: Topics you find harder return sooner in review sessions, spread over days and weeks.

Murray et al. (2025); Emeny & Rohrer (2021) · Source

Mixing topics and matching challenge to the learner helps students choose methods rather than repeat one procedure blindly.

Here: The next question adapts to your mastery and mixes topics instead of giving everyone the same worksheet.

Cognitive science on spacing, interleaving, and adaptive practice

Practising the same skill with new numbers and wording builds understanding instead of answer memorisation.

Here: Questions are original variants generated from exam-style skills — same method, new problem every time.

Platform design based on retrieval and transfer principles