The test was on Sunday. The marks went out Sunday.
A coaching centre lives on its test cycle, and the test cycle is throttled by one thing: how long it takes to check the copies. This removes that constraint without removing the teacher from the marks.
The problem is the gap, not the hours
Faculty losing weekends to copy checking is the complaint. It is not the real cost.
The real cost is the gap between writing a test and getting it back. A student who sees their mistake on Sunday evening, while they still remember what they were thinking, learns from it. The same student, handed the same paper eight days later, looks at the total and puts it in a bag. You have paid for all the teaching either way.
Which is why centres quietly test less than they know they should. Not because tests do not work, but because each one buys a week of checking.
How it fits a centre’s week
- Run the test exactly as you run it now — in the room, on paper, in front of you. Nothing about the test changes; only the checking does.
- Students photograph their own sheets before they leave. Forty students, forty phones, one minute. Nobody has to be chased for a paper afterwards, and your front desk is not collecting bundles.
- Marks and per-question feedback come back the same day. The faculty member reads what was flagged rather than marking from scratch.
- You see what the batch got wrong before the next class, which is the difference between revising a chapter and revising the two questions that actually collapsed.
The volume question, answered before you ask it
A batch of forty sitting a ten-page test is four hundred pages, and phone capture handles that comfortably. If you are running two hundred students through a full thirty-page paper, that is a different job and worth a conversation first — at that volume a sheet-feed scanner beats forty phones, and we will tell you so.
What the owner gets that the teacher does not care about
Faculty want their Sunday back. That sells itself. Two things matter more if you are running the centre:
- Test frequency stops being a cost decision. If checking is no longer a week of faculty time, the number of tests you run is set by what your students need rather than by what your staff can absorb.
- A record of exactly where each student is weak, question by question, across every test — which is the thing you currently reconstruct from memory when a parent asks at the end of term.
What it does not do
It does not replace your faculty’s judgement, and there is no setting that lets it try. Marks are prepared, flagged where they are uncertain, and released only when a teacher approves them. On a paper where the AI could not read a page, it says which page and hands it back rather than marking around it. That is deliberate — the reasoning is here.
If you are not a school-board centre
The checking problem looks different depending on what you are preparing students for, and in one case it barely exists at all. We have written the honest version for each:
- UPSC Mains — where the bottleneck is feedback volume, and where we are explicit about what software cannot judge.
- CA and CS mocks — where the marks live in the working notes.
- JEE and NEET centres — where the objective papers need nothing from us, and the board-pattern tests do.
Preparing students for a board? That is the main case: CBSE, ICSE, or a state board.
Work out what checking costs you now
Four numbers you already know — batch size, tests a month, minutes a paper — give you the hours and the working days a year. The calculator is here, and nothing you type leaves your browser.
Where centres are already asking
The checking problem has a local shape — which board, which medium, how big the sections are. We have written up the places we are asked about most:
Kota · Patna · Indore · Jaipur
Common questions
Is this OMR software or does it grade subjective answers?
Subjective answers. OMR machines already read multiple-choice sheets in minutes and cost almost nothing to run, so there is no reason to put AI anywhere near them. What has not been solved is the written paper — derivations, long answers, diagrams, the marking a teacher does by hand on a Sunday. That is the part this does.
How accurate is AI handwriting recognition on Indian exam papers?
We publish no accuracy percentage, and you should be wary of the ones you see: a figure quoted without a stated test, handwriting sample and reference marker is not a measurement. The answer we can give is the free first paper — your students’ handwriting, your marks beside ours. Anything the system cannot read clearly is handed back to the teacher with the page named, rather than guessed at.
Does it work on Hindi medium answer sheets?
Yes, including the common case of an answer written in Hindi with English technical terms in it. Sections running in both mediums can be marked by the same faculty member against the same marking scheme.
What does a teacher have to do before the first paper is evaluated?
Upload the question paper. The marking scheme is derived from it, so no answer key has to be written out. After that the sheets are photographed — by the students on their own phones, or by staff — and question-wise marks come back for the teacher to approve. No mark reaches a student until a teacher has approved it.
Before you send us anything
It is fair to want to see the output first. A real paper we marked is published in full — every page, every mark, the question that scored zero and the one the grader missed on its first pass.
Try it on last week’s test.
One you have already marked, so you can put our marks beside yours on your own students’ handwriting. About fifteen minutes, and the first one is free.
We reply the same day.