A small team split between engineers and practising teachers. The teachers write the rubrics and overrule the engineers, which is the only arrangement under which an AI marking product should be built.
Not values — six structural facts about the team that explain why the product refuses the things it refuses.
Our curriculum leads have taught to a board syllabus for five years or more. They write every rubric, define what earns partial credit, and hold a veto on any generated paper. It is a staff role with authority, not a quarterly consultation.
Agreement rates between the marker and reviewing teachers are read by the whole team every fortnight — not just by the ML engineers who could explain away a bad number.
Every case study on this site names what got worse before it got better, and which figures we would defend least strongly. A case study with no cost in it is marketing.
Including engineers. It is not optional and it is the single thing most people here say changed how they build. You cannot design a teacher’s review queue from a backlog ticket.
A contractual commitment rather than a setting, written into the privacy policy. It is the question parents ask most, so the answer is fixed rather than configurable.
Built for CBSE, ICSE and state boards, on IST support hours, priced in rupees. Not a Western product with a rupee price list bolted on.
Anyone claiming their AI marks a child’s paper fairly should be willing to say who they are. Credentials matter more here than in most software.
Fill these in before launch. Real names, real credentials and real photographs. Placeholder bios are worse than no team section — they signal the opposite of what this page exists to establish.
[Background — e.g. 8 years in education technology; previously led product at …]
[Credential — e.g. M.Sc. Physics, 12 years teaching CBSE Standards 9–12]
[Background — e.g. previously built ML evaluation systems at …]
[Credential — e.g. 9 years teaching ICSE, state-board paper setter]
Every one has been requested by a paying school, and each refusal costs us money. They are listed here so you can hold us to them.
Not hidden behind a permission — not calculated. Students cannot see one, parents cannot, and neither can a teacher or a head of department. A rank tells a parent who won without telling them anything they can act on, and it is the single largest driver of exam anxiety we are in a position to remove.
Schools have walked away over this. We would rather that than ship it.
A paper reaches students when a named teacher approves it. A result reaches a parent when a named teacher releases it. Every request for an automatic mode has been for convenience, and the cost of being wrong lands on a child rather than on us.
Student answers, scripts and results are never used to train models, and questions a department writes are never shared with another school. It is in the contract rather than in a settings page, so it cannot be changed by a plan upgrade or a policy update.
No upsell in the student interface, no advertising, no data sold to anybody. A child using this at school did not choose the vendor and cannot opt out, which makes marketing to them straightforwardly wrong.
If any of those four ever changes, it changes on this page first, with a date.
A real address and a real person on the other end. Schools checking whether a vendor exists should be able to find out in one click.
[REGISTERED ENTITY NAME]
[REGISTERED ADDRESS]
CIN [CIN] · GSTIN [GSTIN]
[email protected]
+91 80 4000 1234
School hours IST, and through both exam windows.
[email protected]
For access, correction, export or deletion requests. See the privacy policy.
Credentials matter more here than in most software, so these are the ones we get asked hardest.
The team named above, with examiners who have marked board papers doing the rubric work rather than engineers guessing at it. When we say an answer is read against your rubric, the people who designed that reading have sat in front of a stack of forty scripts themselves.
Because it tells a parent who won without telling them anything they can act on, and it is the largest driver of exam anxiety we are in a position to remove. It is not hidden behind a permission — it is not calculated, so there is nothing to switch on. Schools have walked away over this, and we would rather that than ship it.
You do. Papers, rubrics, results and chapter reports export as CSV and PDF whenever you ask, including after you have given notice. It is in the contract rather than left to goodwill, because a school’s exam history should never be leverage in a renewal conversation.
No. Student answers are used to mark that answer and to build that student’s chapter map. They are not pooled into training data, and they do not leave your institute’s scope.
Small enough that the people who wrote the marking logic answer support mail. It matters in both directions: you get a straight answer quickly, and we will tell you plainly when something is not on the roadmap for months.
Book a free 30-minute demo. You will speak to someone who has taught, not only to someone who has sold.