A growing community, a clear gap, and new tools that work
Every figure on this page links to its source. Estimates are marked.
The gap
Muslim children in full-time Islamic schools
~235 schools, ~32,000 students (2011, the latest national study)[6]
Where we sit on price
Families in Santa Clara County already pay an average of $26,301 for private school.[15] AI-first schools such as Alpha charge $75,000 in Palo Alto.[16] We plan $32,000 before aid, ~$24,000 after average aid.
A caution we take seriously
Alpha's Arizona charter scored well below its targets.[19] Technology alone is not enough. Our answer: guides in every room, in-person Qur'an teachers and small-group tutoring in school hours.
What the evidence says, and what we do about it
More than double the learning gains
A well-designed AI tutor more than doubled learning gains compared with an active-learning class (effect sizes 0.73–1.3 SD) in a study of 194 Harvard students.[27]
What we do: Design matters: the tutor followed good teaching practice.
Hints, not answers
With a plain chatbot, practice scores rose 48% but exam scores fell 17%. A hint-based tutor raised practice 127% with no exam harm (~1,000 students).[28]
What we do: Mizan's tutor gives hints, never answers.
AI with a teacher in the room
AI sessions with a teacher present produced +0.3 SD in six weeks.[29]
What we do: Guides stay in the room, every session.
AI that coaches the adult
AI coaching of human tutors raised mastery 4 percentage points, and 9 points for weaker tutors.[30]
What we do: Mizan coaches guides, not just students.
Tutoring works best in school hours
Across 89 trials, human tutoring added +0.29 SD, strongest during school hours and in early grades.[31]
What we do: Small-group time is built into the school day.
Rooms shape learning
Classroom design explained 16% of the variation in learning progress (3,766 pupils, 153 classrooms).[32]
What we do: Daylight, fresh air and calm walls in every room.