Claude
Azure AI
Cloudinary
Author the quiz.
Thirteen question types, mathematical notation, images and video. Build a bank once and reuse it across every quiz that needs it.
Every quiz, start to finish.
Author it, sit it, mark it, report it. One record throughout.
ClaudeAzure AICloudinaryPaystackmNotify
One record, every screen.
Questions, candidates, marks and money share a single tenant-aware database.
There is no nightly import and no reconciliation spreadsheet. A mark entered on the results screen is the same mark the report card prints, the API returns and the audit log remembers.
- 14
- Question types, from multiple choice to ordered lists and free text
- 39
- Endpoints on the public API, each behind a scoped key
- 2
- Delivery channels for an invitation: email, SMS, or both
- 1
- Source of truth, scoped by organisation and by location
Send the invitation.
Invite by email, by SMS, or by both. A candidate with no number is downgraded to email and named, never quietly dropped.
Mark what a machine can.
Objective questions score themselves the moment a quiz is submitted. Free text waits for a human, and the total says so rather than counting it as zero.
Send it home.
Terminal report cards, certificates and fee statements, generated from the marks already on the record and frozen the moment they are published.
A school runs on terms, not quizzes.
The School Examination add-on turns the platform into something a headteacher recognises.
- Groups, courses and an academic calendar. Years and terms are organisation-wide, and the term after a year’s last term is the first term of the next year.
- A timetable with rooms and invigilators. Clashes are refused on creation and on every later move, including the ones that put an invigilator in two rooms at once.
- Continuous assessment, weighted. Assessment score and examination score combine on a weighting you set, and the two halves have to total one hundred.
- Report cards that freeze. Publishing writes a revision and never edits the last one, so a reprint matches the copy a parent is already holding.
- Promotion that keeps the past intact. Moving a group closes one membership and opens another, so last year’s results stay attached to last year’s group.
Photograph the past paper.
A document you already have is a source, not a blank page.
- Upload a photograph or a PDF. The text is read once and kept, so reopening a document costs nothing and never re-reads a page you have already paid for.
- A paper that already has questions is not generated. They are pulled straight out of the text with no model involved, because paying to have printed questions reinvented helps nobody.
- Nothing is written without you. Everything read or generated arrives as a proposal on a review screen. You accept it, edit it, or throw it away.
- Corrections sit beside the read. Editing the extracted text never overwrites what the reader produced, so the original stays available.
A public API, and a reference that cannot go stale.
Thirty-nine endpoints, scoped keys and outbound webhooks.
The OpenAPI document and the router are asserted against each other in both directions. A route missing from the specification fails the build, and so does a documented route that no longer exists.
- A key’s scope narrows the key. The service user’s permissions narrow the user. What a request may do is the intersection.
- A webhook subscription can never reach further than the key that created it.
- The published reference is generated from the same document the tests assert against.
# List the quizzes a key is scoped to read
curl https://quizmaster.clickerapps.com/api/v1/quizzes \
-H "Authorization: Bearer qm_live_..."
# 403 if the key does not hold quizzes.read.
# 403 too if its service user lacks the permission.Built to be checked.
Four promises, each of them enforced rather than asserted.
A published result is frozen.
Re-publishing writes the next revision and leaves the last one untouched. A position in the group is derived from every other child in the room, so one correction would otherwise restate a document that has already gone home.
It keeps working when the line drops.
The application shell and the screens you were last on are cached, and writes queue locally and replay in order once the connection returns.
Every change is on the record.
Each mutation runs through an audited stored procedure carrying the actor, the organisation and the location. Nothing writes to the database around it.
A result names one candidate and never another.
What a participant may be told about their own attempt is decided in one place. A position is a number out of a count, and a classmate is never named on it.
Start free. Stay free if it fits.
No card, no trial clock, no feature held back to force an upgrade.
Free
GHS 0 no time limit
Everything you need to run real examinations for a single organisation.
- Up to 3 quizzes, 60 questions and 50 learners
- All thirteen question types
- Email and SMS invitations, automatic marking
- Certificates and result exports
- Offline-first, on every screen
Pro
GHS 250 per month, billed monthly
For groups running examinations across more than one campus.
- Everything in Free, with no limit on quizzes, questions, learners or staff
- Multiple locations under one organisation
- Up to 50 AI-written questions at a time
- The School Examination add-on, available on request
Questions, answered plainly.
Can we try it properly before paying anything?
Yes. The free plan has no time limit and asks for no card. It is the same software, not a demonstration, and the work you do on it stays yours if you later upgrade.
What happens when the internet drops mid-examination?
The screens keep working. Writes queue on the device and replay in order once the connection returns, and the candidate sees the quiz rather than an error page.
A mark was wrong and the report cards are already out. Now what?
Correct the mark and publish again. That writes a new revision and leaves the issued one intact, so you can see exactly what each family was sent and when it was superseded.
Can we run more than one campus?
Yes, on Pro. Locations sit under one organisation, results roll up across all of them, and a member of staff sees only the campuses they have been given.
Is there an API we can build against?
Yes. Thirty-nine endpoints under /api/v1, authenticated with scoped keys, plus outbound webhooks. The full reference is published at /developers/api and needs no account to read.
Who can see a candidate’s results?
Only the people you grant it to, and the candidate themselves. What a participant is shown about their own attempt is decided in one place in the code rather than per screen, and it never names anybody else.
Set the quiz. Send it home.