The learning engine
The algorithm decides what a child knows. The AI only writes.
Anything that must be correct, repeatable and explainable to a parent runs as an algorithm. The model drafts and explains — and a person reviews it before a learner ever sees it.
“AI-powered” is doing an enormous amount of work on most edtech homepages.
A model that guesses a child's mastery will one day tell a parent their child is weak at fractions on the strength of four questions and a bad afternoon. That is not a technical problem — it is a thing you cannot explain to a family, which means it must not be the thing that decides.
What that looks like in practice
Mastery is statistics
What a learner has and has not understood is computed by an explainable model a teacher can reason about. No verified production system of any scale uses deep knowledge tracing for this — not Khan Academy, not the large assessment vendors — and neither do we.
Every item is reviewed by a person
Explanations and questions are drafted offline, checked by a human reviewer, and only then published. Nothing reaches a learner unreviewed.
Difficulty is calibrated by real answers
Not assigned when a question is written. A model cannot judge how hard a question is for a child — the evidence for that is unambiguous — so live responses decide it.
Revision fights forgetting
Topics return for review as they are about to fade, rather than on a fixed weekly cycle.
Bands, never a raw probability
Not started, learning, practising, confident, mastered, with a trend. A parent can act on that.
Safety is fixed, not improvised
A learner in distress meets a tested response and a named human contact. A model does not get to improvise that moment.
What the school sees, and what it does not.
Practice a school sets is schoolwork: the teacher sees progress by topic, as they would see an exercise book.
What a family buys for itself is a different matter. The school has no view of it — not the purchase, not the content, not the usage. That separation is built into the data model, which is the only place a promise like this can actually live.
The questions worth asking
Including the ones where our answer is no.
- Which subjects and classes are live?
- Content is built subject by subject. We will tell you exactly what exists for your child's class at a demo, and we will not list a subject that is not there.
- Is my child's data used to train a model?
- No.
- Can a school switch the tutor off?
- Yes, entirely, or for particular classes.
- Does it replace a teacher?
- No, and any product that tells you it does is selling you something. It finds gaps and drills them. Teaching is the other thing.
Instead of a wall of logos
What you can check for yourself.
Your data never leaves India
Hosted in Mumbai. Every institution's records are separated inside the database itself — not by a line of application code somebody could forget to write.
Built for the DPDP Act 2023
Parental consent, withdrawal and erasure are working features, including the stricter rules that apply to a child's data.
UDISE+ without the fortnight
Your return is generated from the records you already keep, and every figure can be opened up to show the rows it came from.
Readable by everyone
Measured against WCAG 2.2 AA across every screen on every release — because a grandparent at the school gate uses this too.
Ask us the hard question at the demo.
Ask what happens when the algorithm is wrong about a child. Any vendor worth buying from has an answer they have thought about.