Quick reading and maths level checks that sort children into catch-up groups
TL;DR Levelsort helps primary teachers find out, child by child, who can read a short story or do a subtraction sum, and then teach in groups by level for part of the day. CSR foundations, NGOs and low-fee private schools pay for the checks and the plan that follows.
Most children in Indian primary classes are taught at their grade level, whatever level they've actually reached. ASER 2024 found that only 23.4% of Class 3 children in government schools could read a Class 2 text, and that's the best figure since the survey began[1]. By Class 5, 44.8% of government-school children could read that text against 59.3% in private schools[1].
That gap is where inequity grows. A teacher with 40 children and one textbook rarely knows which eight can't yet read words, and the annual report card doesn't tell her in time to do anything about it.
What it does, in regulatory terms: learning-level assessment by teachers, with no psychological or behavioural profiling. The data is children's personal data under the DPDP Rules 2025, which need verifiable parental consent unless an education exemption applies[2].
Every six to eight weeks the teacher sits with each child for about five minutes. The child reads letters, words, a paragraph and a short story aloud and solves a few number tasks. Speech recognition suggests where the reading broke down and the teacher confirms with one tap. The app then sorts the class into four or five level groups and gives each group a set of daily activities for a one-hour block. Results can be copied into the learning sections of the school's Holistic Progress Card[3].
I reframed the seed, which pitched AI-powered holistic understanding of each student. Profiling children's personalities from data is risky and hard to validate. Knowing their reading and maths level and grouping them by it has strong evidence behind it - one large literacy NGO's level-based learning camps in Uttar Pradesh doubled the share of children who could read a paragraph or story[4].
Teachers and children use it. CSR foundations and NGOs that run catch-up programmes pay per child per year, because they must report learning gains to funders. Low-fee private schools are a second buyer, and state contracts are possible later but slow. I found no public prices for comparable tools, so the fee is my assumption to test.
NEP 2020 asked for a holistic progress card and suggested software to track student growth[5]. NCERT's PARAKH has published stage-wise Holistic Progress Cards and launched a digital HPC portal[3]. One AI non-profit reports more than 15 million AI oral reading assessments across two states as of January 2026[6], so the speech piece already works at scale in Indian classrooms.
There is no public figure for what Indian buyers spend on learning-level assessment. Adjacent signals show demand though. One edtech reports its adaptive learning labs in more than 5,300 locations across 16 states[7], and the AI non-profit's 15 million reading checks were run with state governments[6].
Treat both as proof that buyers exist, not as your market size.
AI non-profits build oral reading fluency checks for state governments[6]. Big tech offers a free reading-practice app that works offline[8]. State-focused edtechs run chatbot learning and adaptive labs with states[7]. Teaching at the Right Level, the level-grouping method from a large literacy NGO, is the method itself, largely run by NGOs and governments[4].
Free state tools cover a lot of government schools. Low-fee private schools and CSR programmes outside those deployments are the gap.
| Difficulty | Moderate (45/100) |
| Starting capital | ₹50–90 L |
| Time to launch | 6–9 months |
Children's speech in noisy classrooms and in several languages is hard to recognise, which is why the teacher confirms each result. The bigger challenge is behaviour, because teachers must actually regroup children every day and that takes training and follow-up visits, not just an app.
No licence is needed for a learning-level tool, but working with children and government schools brings permissions.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't trust AI scores for children who speak a dialect or have a speech difficulty without a teacher's check. Never use results to label or hold back a child.
Want to work out the AI layer for Levelsort? Talk to Yash →
I like this for a founder from a foundational-literacy NGO or a state FLN programme who can sell to CSR funders. If you're hoping for fast consumer growth then look elsewhere. Prove one thing first - in 30 classrooms, teachers keep running the level hour for a full term and the share of children reading a story goes up.
Plan for roughly ₹50–90 L, mostly on speech technology in local languages, field pilots and work with education departments.
Yes, usually written permission from the district or state education department. You also need verifiable parental consent under the DPDP Act unless an exemption applies.
Yes, many countries have children behind grade level. Check the local rules on children's data and how schools there approve outside tools.
The free Levelsort validation kit has the line-by-line capital breakdown, the effort scoring, the child-data and progress-card checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for funders and school leaders, and 5 assumptions to test. Download the kit →
Thinking about building Levelsort or something close to it? The make-or-break bit is getting teachers to regroup children every day and then proving learning gains a funder will pay for, which is the kind of plan I help founders put together. 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.