Teaching a child you can actually see.
Teaching, marking and study, for schools and learners.
On the phone the household already owns · Prices in pula, rand and dollars · Built by OrionX in Botswana
Botswana’s reading number dates from 2011. A whole school generation has passed through unmeasured.
One study. Three rows. The third is the one every lesson depends on.
The World Bank observed teaching directly across seven African countries — close to 40% of the region’s population. Not a survey of opinions: observers in the room, tests on the desk.
Source: World Bank, direct observation across seven African countries — Kenya, Nigeria, Mozambique, Senegal, Tanzania, Togo and Uganda.
In maths, 68% of teachers cleared the minimum subject-knowledge bar; in language, only 7% did. But on assessing students — finding out where each child actually is — the number was zero percent, in every country where it was measured. Only 17% could mark writing against a scheme.
This is not a verdict on teachers: 90% of those tested hold secondary qualifications or higher. It is a verdict on their tools. The system trained them to deliver a lesson and never handed them an instrument for the one fact a lesson depends on.
could mark a piece of writing against a scheme
World Bankof teachers tested hold secondary qualifications or higher
World BankA teacher can tell you which children in the room are struggling. Almost none can tell you which specific skill is missing in which child, this week, against the curriculum. That is not a character flaw. It is a missing tool — and a missing tool is a buildable problem.
A system that spends well and cannot see each child does not need more content. It needs sight.
spent per primary-age child in Botswana — 172% above the sub-Saharan average
World Bankof children at late primary age cannot read proficiently, counting those who never got there
World Bank · UNESCO Institute for Statisticslearning-adjusted years of schooling banked by a child in Botswana
World Bankof actual teaching a day, against 5h27 scheduled — and one classroom in three held students but no teacher
World BankBotswana has textbooks, buildings and qualified staff. Time leaks the same way the money does: just over half the timetable happens. The other half is not badly taught — it is not taught.
There is a harder version of this. Botswana’s reading figure derives from PIRLS 2011, and the country runs no national large-scale assessment at the end of primary school.
That last tick is the product. No national large-scale assessment at the end of primary (World Bank).
Botswana ran the trial itself. The finding was not that phones teach — it was that aim does.
Same handset. Same messages. One pitched at each child’s level, one not. Only the aim changed — which is why the answer is a diagnostic instrument and not a device programme.
Measured during pandemic school closures (NBER) · drag to orbit · download as OBJ or GLB
The fix with the strongest evidence on earth is teaching each child at their actual level.
After sifting more than 13,000 research papers, the Global Education Evidence Advisory Panel named just three “Great Buys” — and teaching at the right level is one of them. Buying computer hardware was formally rated a Bad Buy. That last one is worth sitting with, because it is the thing most education technology is actually selling.
Botswana ran the decisive trial itself: during the pandemic school closures, phone-and-SMS tutoring pitched at each child’s level produced gains of 0.12–0.17 standard deviations at $19 per child — comparable to a US programme costing $2,500 per child, and up to 1.1 learning-adjusted years of schooling per $100. The same SMS messages, untargeted, taught nothing on their own.
Sibling, not moduleGreyEd — the education specialist from OrionX, Uhuru’s builder — is built on exactly that finding: short diagnostic checks, a live map of what each learner can and cannot yet do, and practice pitched at that level. It is served at greyed.org. The instrument the seven-country study found missing from every staffroom.
Around it sits the Uhuru platform. Teachers draft structured lesson plans and marking schemes in U Docs — structured pedagogy is another of the three Great Buys, and in Kenya it produced reading gains worth more than a full extra year of learning. Uhuru Chat carries the work on the phones households already own, Uhuru Voice reads and listens for learners who work better by ear or on a thin connection, and Polymath 4.1 keeps everyday use affordable. “Uhuru” means freedom — here it means a teacher freed from guessing.
The same five steps. The difference is entirely in the last two.
The teacher’s week today
Ends in a percentage, and a guessThe teacher’s week with Uhuru
Ends in evidence, weeklyThe teacher’s judgment stays in charge at every node. Uhuru marks against the scheme; the teacher reviews and signs off.
Eight things a staffroom gets that it has never had.
See every child
GreyEd runs short diagnostic checks and keeps a live map of where each learner stands — the instrument the World Bank's seven-country survey found no staffroom had been given.
Teach at the right level
Practice and explanation at the child's level, not the grade's — the approach the global evidence panel names a Great Buy, and Botswana's own trial showed is the difference between learning and nothing.
Mark against a scheme
Upload a scheme and scripts; get strengths-and-weaknesses feedback per learner for the teacher to review and sign. In the African survey, only 17% of teachers could do this unaided — for lack of tooling, not ability.
Prepare structured lessons
U Docs drafts structured lesson plans, worksheets and progress records. In Kenya, structured pedagogy delivered reading gains worth more than one full extra year of learning.
Reach the phone that exists
Built for modest handsets and thin connections — sub-Saharan Africa's 527 million mobile subscribers are a wider learning channel than its broadband.
Track the progression, not the score
A score says a child got 54%. A progression says which skill stalled, in which week, and what comes next. Just over 12% of teachers in the seven-country study could monitor a progression unaided.
Answer in the language the child thinks in
Uhuru Chat and Uhuru Voice work in the language a learner actually uses, then move the same understanding into the language the exam is written in. Nearly 37% of students in low- and middle-income countries are taught in a language they do not understand.
The school keeps its own record
Every plan, script, scheme and learner map is stored in Uhuru Cloud. A teacher who moves schools does not take the term's material with her, and the child's progression is still legible to whoever teaches them next January.
Eight lines a head of department can check against her own term.
Lore lo ojwa lo sa le metsi.
A branch is shaped while it is still wet.
Wood takes a shape only while the moisture is in it. The phone in a parent’s pocket can now say, this week, whether a child can read — not a national report published years after the wood has set.
The teacher standing in front of sixty learners with no instrument for knowing where any of them are.
And the learner whose classroom, one hour in three, is an empty room.
Sixty-two on the register, four subjects, and one term-end mark sheet standing between her and any idea of who is actually behind.
Marks against a scheme when he has the hours, guesses when he does not, and has never once had a per-child picture of which writing skill his classes are missing.
Thinks in Setswana, is examined in English, and answers questions in an evening on her mother’s phone that she would not put a hand up for in the room.
Get the instrument the staffroom was never given.
Short diagnostic checks, a live map per learner, practice pitched at that level. On the phone the household already owns.