of Class Three children in rural India cannot read a Class Two-level text.
Source: ASER, PrathamWhy Suno works
A careful reading loop: teachers decide, and evidence stays visible.
The problem
of Class Five children cannot do division.
Source: ASER, PrathamWhat the ASER figures show
Why reading is the hinge
Early reading opens access to every later subject. When children get far behind, the gap can compound; this is the Matthew effect described by Stanovich.
Reading paths can diverge
Teaching at the Right Level
Children learn best from the skill they are ready to practise, not from a grade label alone. This is the central insight behind TaRL, an approach J-PAL describes as rigorously validated and cost-effective.
Targeted teaching, clearly explained
The loop
- AssessListen to one short reading.
- UnderstandReview teacher-confirmed evidence.
- PractiseChoose one small, useful next step.
- Re-checkSee what changed, with the teacher in charge.
Why the AI is guardrailed
AI drafts word-level observations; teachers confirm the result before it can guide instruction. A PNAS study found unrestricted chatbot help increased practice but exam scores were 17% lower when the help was removed; a guardrailed version removed that harm.
Source: PNASWhat it does for teachers
Teachers using AI report saving about 5.9 hours/week; Suno uses that promise carefully, with teacher confirmation rather than automatic decisions.
Source: GallupNumeracy — in development
Feature in active development
Today's numeracy checks help teachers notice addition and subtraction patterns, then review them before making a practice plan. They are not presented as a validated learning intervention.
ASER reports 69.3% of Class Five children cannot do division.
Source: ASER, Pratham