Back to Bigger Picture

BIGGER PICTURE Learning & work

A maths tutor that fits inside a WhatsApp conversation

A computer lab is not the only route to personalised practice. Rori brings short maths lessons, questions and hints to basic mobile devices over modest data connections.

AI-generated illustration: A smartphone and exercise book joined by a mathematics practice session.
A smartphone and exercise book joined by a mathematics practice session. Conceptual illustration; not a photograph or an exact depiction of the real project.AI-generated illustration of the idea, not a photograph of the actual people, equipment or event.
School routine
Two 30-minute sessions weekly
Observed score growth
5.13 vs 2.12 points out of 35
Evidence boundary
Preliminary study; 11 schools

How the idea works

  1. Choose a small maths skill

    Short curriculum units match practice to the learner.

  2. Answer, then receive a hint

    Conversational feedback precedes a worked solution.

  3. Check learning outside the chat

    School assessments compare progress with regular instruction.

Start where the student is

Rising Academies built Rori for settings where computers and reliable broadband are scarce. Its mathematics curriculum is divided into small lessons. Students answer in conversational language; if they struggle, the tutor gives a hint before showing a worked solution. The goal is repeated practice with feedback that an overstretched classroom may struggle to provide individually.

Sources: [1]

An hour a week, inside the school day

In the Ghana study, participating schools were assigned to intervention or comparison groups. The Rori group used phones for two half-hour study sessions a week alongside regular classes, with teachers present for technical help. This was a supported school routine, not proof that an unsupervised chatbot works equally well in every home.

Sources: [1]

The result is promising—and preliminary

The preliminary study reported gains of 5.13 points with Rori versus 2.12 without it on a 35-question maths assessment. Of 637 students assessed at baseline, 477 had paired results; 160 lacked a follow-up assessment. Those missing results matter when interpreting the comparison. The study establishes a promising result in these schools, with substantial uncertainty about who is represented in the final sample.

Sources: [1]

The next test is ordinary public schools

Rising Academies subsequently announced a larger randomized study across 100 Ghanaian public schools. That is a test of whether the approach can travel beyond its early setting, not an established additional success. The inspiring idea remains practical: design individual feedback around the devices and connectivity learners can actually access, then test the whole delivery model.

Sources: [2]

The original sources

  1. Henkel and colleaguesEffective and Scalable Math Support: experimental manuscript, version 2
  2. Rising AcademiesTesting Rori at scale: a new RCT in Ghana

What to explore next

The tutor gets a coach at the difficult momentCan a hospital spot an empty appointment before it happens?
Explore all 99 applications