Blank page to declared result,
in one sitting.
Six steps, and the examiner stays in control of every one. The AI drafts and marks — it never decides on its own what the paper should be, and there is no step where it publishes something a teacher has not read.
What happens,
and who decides it.
The homepage version of this is the summary. This is the same six steps with what you control, what the platform does, and what it refuses to do at each one.
Choose the syllabus
Standard, subject and the chapters in scope. Upload your own chapter PDFs or notes and the paper is drawn from those instead of a generic bank.
Scope is a hard limit, not a preference — a question from Chapter 9 cannot appear in a paper you scoped to Chapters 4 to 7.
- You decide which chapters, and what to upload
- It does tag every question to a chapter as it drafts
- It will not reach outside the chapters you picked
Pick the question types
Section by section, from a palette of twenty-plus formats: ten MCQs, five fill in the blanks, three short answers, two long. Set marks per question and the total duration.
The AI fills the shape you defined. If your chapters cannot fill it, you get told which section fell short and by how much rather than a quietly short paper.
- You decide the marks scheme and the difficulty ratio
- It does fill the blueprint and balance across chapters
- It will not hand you a 47-mark paper without saying so
Review and edit
Every generated question is editable, swappable or removable, and each arrives with its answer key and rubric attached. Type your own question straight into any section and it counts toward the blueprint.
Nothing reaches a student unread. There is no auto-publish switch, and we are not going to add one.
- You decide what goes out, question by question
- It does supply a key and rubric with every question
- It will not publish anything you have not approved
Schedule and assign
Send it to Standard 10-B, to a whole year group, or to one student sitting a re-test. Set the window and the duration, and choose whether questions shuffle per candidate.
One paper generated once can be assigned to twelve divisions, which is what makes comparing them afterwards a fair question.
- You decide who sits it, and when
- It does shuffle per candidate so neighbours differ
- It will not re-set a question a division has already seen
Students attempt it
On a phone, tablet or lab machine. Answers save on the device as they type and sync when the line returns, so a dropped connection costs the seconds it lasted and nothing more.
The invigilator sees one screen for the whole room: who is in progress, who has submitted, who never signed in.
- You decide when to grant extra time — it is logged
- It does keep working offline and sync in order
- It will not lose an attempt to a bad connection
Results and analytics
Objective sections are marked on submission. Written answers are read against the rubric you set, with step marks for partial working and the phrase that earned each point quoted back.
You release the result. The moment you do, every dashboard updates at once — student, teacher, institute and parent — broken down to the chapter.
- You decide every final score, and when it is released
- It does mark, explain each deduction and build the chapter map
- It will not compute a rank — there is no setting for one
Steps one to four take a teacher about ten minutes in total. Step six takes under a minute. Step five is the exam.
Three columns,
and one of them is empty.
The third column is the point of this page. Pick a column to read it on its own.
A rank is not a permission we withhold. It is a number the platform does not compute, so there is nothing to switch on.
One paper, forty scripts,
start to declared result.
A fortnightly Standard 9 science paper for one division. Scored on how much of it a teacher has to do personally. Click a row.
Four things this
process will not do.
Constraints, not missing features. Each of them is a decision we would have to reverse deliberately.
Go outside the chapters you pickedScope is a hard limit, not a preference
If Chapters 4 to 7 cannot fill your blueprint, you are told which section fell short and by how much. It does not borrow a question from Chapter 8 to make the total add up, and there is no setting that loosens this. A paper that quietly tests something you did not teach is worse than a paper that is short.
Publish without a teacherThere is no auto-publish switch
And we are not going to add one. A paper reaches students when a named teacher approves it, and that approval is logged with their name and a timestamp. The same applies to results: marking finishes, then it waits for you. Every request we have had for an automatic mode has been for convenience, and the cost of being wrong lands on a student rather than on us.
Re-set a question a class has already seenUnless you ask for a repeat
Every question a division has been shown is remembered, and repeats are excluded from that division’s future papers automatically. You can deliberately ask for a repeat — re-testing the same chapter is often exactly right — but it will not happen by accident because a bank was small that week.
Compute a position in the classThere is no setting for one
Not hidden behind a permission — not calculated. Students cannot see a rank, parents cannot, and neither can a teacher or a head of department. What replaces it is a chapter map, which is actionable and much harder to feel humiliated by. This is the constraint we get asked to remove most often, and the one we are least likely to.
If any of those four would be a problem for how your school runs exams, it is worth raising on the first call.
What people ask
about the process.
The five that come up almost every time.
Longer than the ones after it, and the difference is not the software. Setting and generating is about ten minutes from the first time. What takes longer is agreeing your marks scheme and writing down what earns partial credit on a written answer, because most departments have never put that in writing. Budget an afternoon of meetings for that once, and every paper afterwards is the ten minutes.
Yes, and it changes the output substantially. Upload chapter PDFs, notes or a scanned textbook section and questions are drawn from those rather than from a generic bank, which matters most for state boards and for schools teaching a syllabus with local content. Anything you upload stays yours and is not shared with other schools.
Steps one to four and step six all still work. Papers print to A4 with the marking scheme, students sit them on paper, and you either enter marks per question or upload scanned scripts for marking. You lose the automatic marking of written answers when handwriting is poor, which is a real cost — but the generation, the chapter tagging and the analytics are unaffected.
The teacher who set it, and anyone they share it with. Not students, not parents, and not other divisions’ teachers unless you give them access. Papers are held unpublished until the scheduled window opens, and the question order is shuffled per candidate at the moment they sit it rather than stored per student.
For objective sections, immediate — they are marked as the paper closes. For written answers across forty scripts, typically thirty to ninety seconds depending on length. That is the marking. The result reaches students when you release it, which is a separate decision and can be minutes or days later.
Try it on one class.
Book a free 30-minute demo. Bring a paper you already ran and we’ll take it through all six steps live, so you can see exactly where your ten minutes go.