A different maths sheet for every child, printed from one teacher's phone
TL;DR Mathsheet brings personalised practice to low-fee private schools that have no tablets. The teacher photographs a class's short answer sheets on one phone, and the next morning each child gets a printed worksheet pitched at their own gaps. Schools pay per pupil per year.
Most Indian primary children are still behind in basic maths. In ASER's 2024 rural survey, only 33.7% of Std III children could do a two-digit subtraction, and 30.7% of Std V children could solve a simple division[1]. Those are the best figures in over a decade, and they're still low.
A class of 40 usually spans three or four grade levels of ability, and one teacher with one textbook teaches to the middle. Government schools enrol 66.8% of rural children aged 6–14, and most of the rest attend private schools[1]. A lot of these are low-fee schools with no computer lab and no tablets per child.
Mathsheet starts with foundational maths for Std 3–5. Twice a week, children do a short diagnostic sheet in class. The teacher photographs the sheets with a phone, and the app reads the answers, works out which skills each child has and hasn't mastered and generates a personal practice sheet for the next day. The school prints them on its own printer or at a local print shop. The teacher gets a one-page class view showing which five children need help with borrowing and which ten are ready to move on.
It's paper in and paper out, so no child needs a device and screen time stays at zero. Sheets come in Hindi and the state language as well as English.
The seed pitched personalised content delivery in general. That's a crowded space, so I narrowed it to the schools the leading products don't reach - low-fee private schools without devices.
The school owner pays, usually folding a small amount into annual fees or a separate activity fee. My working assumption is a per-pupil yearly fee low enough for schools charging a few hundred rupees a month, and I found no public data on what these schools spend on learning products. Parents are the indirect payers and need to see the sheets coming home.
The evidence for personalised practice is strong. In a Delhi trial, children given access to an adaptive learning platform scored 0.37 standard deviations higher in maths after 4.5 months, and weaker students gained the most relative to where they started[2]. Phone cameras and handwriting recognition are now good enough to read a child's digits reliably, which wasn't true five years ago. Government schools are getting PAL labs[3], which leaves low-fee private schools further behind.
There is no public figure for low-fee private school spending on learning products. The narrowest proxy - one edtech runs its adaptive learning product in more than 5,300 PAL labs across 16 states[3], which shows that states pay for personalised practice when it reaches children.
ASER's enrolment split, with roughly a third of rural children outside government schools[1], gives you the reach and not the market.
An established Indian adaptive-learning platform offers online practice in maths, science and English[4] and is the product behind the Delhi trial. A state-focused edtech sells a personalised-learning app in nine languages, mainly through state governments[3]. Both need a screen for each child or a lab.
Print-based adaptive practice for schools without devices is the gap I see.
| Difficulty | Moderate (45/100) |
| Starting capital | ₹55–95 L |
| Time to launch | 6–9 months |
Reading handwritten digits from phone photos in poor classroom light is harder than it sounds, and it has to be close to perfect because a misread answer sends a child the wrong sheet. Building a skill map and item bank for each grade and language takes months. Selling to hundreds of small school owners one by one is slow.
No special licence. Practice material needs no board approval, so you only need the baseline, plus care with children's data.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't trust the reader blindly. A misread digit sends a child the wrong sheet, so show the teacher anything the software isn't sure about and keep a human check on the item bank.
Want to work out the AI layer for Mathsheet? Talk to Yash →
I like this for a founder who knows low-fee private schools from the inside and can build strong computer vision. If you want a consumer app then look elsewhere. Prove one thing first - teachers in five schools keep scanning twice a week for a full term without being chased.
Plan for roughly ₹55–95 L, mostly on handwriting recognition, the skill map and item bank in each language, and field sales to schools.
No board approval is needed for practice material. You need GST once over the threshold and DPDP-compliant consent for children's data, collected through the school.
Yes, anywhere low-fee schools lack devices but teachers have phones. Rebuild the item bank around the local curriculum and check your country's child-data law.
The free Mathsheet validation kit has the line-by-line capital breakdown, the effort scoring, the child-data checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for school owners and teachers, and 5 assumptions to test. Download the kit →
Thinking about building Mathsheet or something close to it? The make-or-break bit is getting busy teachers to scan sheets week after week, and that's the kind of habit problem I help founders work through. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.