FAQ
Frequently asked questions
Answers on pricing, limits, AI and integrity, authoring questions, proctoring, and getting things running. If yours is not here, ask us — we would rather answer it than have you guess.
Pricing
What it costs and how it is counted
Full detail — including a cost calculator — is on the Pricing page.
You pay per assessment — ₹90 in India, $1.00 elsewhere — and nothing else. There
are no seat licences and no annual minimum. Start with a free trial: 3 months or
50 assessments, whichever comes first, with no card required.
See pricing.
One assessment that has not been cancelled. An individual assessment counts once;
a bulk batch counts once per participant it creates. Cancel an assessment and it
stops counting. Usage is calculated live from your assessments — there is no
wallet balance to reconcile.
Your account stays exactly as it is — quizzes, participants, and past results are
untouched. You simply cannot create new assessments beyond your allowance until
you buy credits. The trial ends at 3 months from signup or 50 assessments,
whichever comes first.
It still counts, because the assessment was created. Cancel it and it stops
counting — cancellation is what frees the credit, not the participant's no-show.
No. Add as many ADMIN and STANDARD users to your organisation as you need — the
only thing you pay for is assessments.
No. Your currency is set by the country you choose at signup — India bills in INR,
everywhere else in USD — and it is fixed for the account after that.
Every purchase generates an order with tax computed and itemised at the time of
purchase, against your organisation's billing profile. The per-assessment rate is
recorded on the order, so a later price change never rewrites what you already
bought.
No. Credits are non-refundable once purchased, and they are valid for
2 years from the date of purchase — credits unused at the end of
that period lapse. Because you buy them in whatever quantity you need rather than
committing to a plan, we suggest buying in line with your near-term need. See our
Terms of Use.
Payments are processed by Razorpay. Card details are handled by the payment
provider and are never stored by ZiniApps — see our
Security page.
Limits
How much fits in a quiz, a batch, and an account
These are the defaults every account starts on. They are set on our side, not built into the product — if your exam needs more room, tell us what you need.
By default a quiz holds up to 720 questions across
20 sections, and can run for up to 3 hours.
Each participant is served up to 360 questions from that pool.
The two numbers are different on purpose: you keep a large bank of questions, and
each sitting draws a shorter paper from it, so no two participants get quite the
same set. How quizzes work.
Up to 1,000 participants in a single bulk upload by default. That
is a limit on one upload, not on your account — run several batches and there is no
ceiling on how many people you assess in total. Each participant in a batch is one
billable assessment. See pricing.
By default, 50 quizzes and 100 team accounts per
organisation. Team accounts are not charged for — pricing is per assessment, so
adding a reviewer costs nothing. See pricing.
A question carries up to 5 answer choices (four is the norm), and the question
text and each choice can each run to 2,000 characters — enough for
a full case-study prompt. Questions may include mathematical notation, and can
optionally reference an image by
https:// URL. Questions are imported
as structured JSON, so anything over a limit is reported with the question number
at import time rather than surfacing mid-exam.
Yes. Every number on this page is a setting we control, not a hard boundary in the
product — they are sized so a normal account never meets them. If your exam is
larger than the defaults allow, raise a ticket and tell
us the shape of it. There is no upgrade tier to buy: pricing stays per assessment.
AI & integrity
What stops AI assistance, and what does not
Full detail is on the Anti-AI measures page.
Several things, and we will not claim they add up to prevention. Every attempt
draws its own paper with its own answer order, so a leaked question set is worth
little. Questions are masked and revealed one block at a time, which makes a
single screenshot useless. The paper is fetched one question at a time, so the
whole exam is never sitting in the browser. And correct answers are never sent
to the exam device at all.
See all the measures.
The question and each answer choice start hidden. Hovering the pointer over a
block reveals that one block; moving away re-hides it. Only ever one block is
legible, so a screenshot captures a fragment rather than a question — anyone
wanting the whole thing has to go back and capture it piece by piece.
No. There is no setting and no plan tier for it. Masking, the per-attempt paper,
and the answer shuffle apply to every exam and every attempt, including on the
free trial.
Not conclusively, and we do not sell a detector. What we have is answer-timing
analysis computed on our servers from the stored answers — useful because it
needs nothing from the participant's machine and so survives a tampered exam
client. It is a review input for a person, not a verdict. We do
not produce an "AI likelihood" score, because the evidence does not support a
single authoritative-looking number.
No — and this is worth knowing before you choose us. The reveal gesture needs a
mouse or a pen; there is no touch, keyboard, or screen-reader path. Touch-only
devices are stopped at the pre-check, before the timer starts, so nobody
loses an attempt to it. If your participants will be on tablets or you need a
screen-reader accommodation, talk to us first.
Authoring
Getting your questions in
The full field reference and downloadable samples are on the Product page.
Proctoring
What is watched, and what that is worth
The full account, including the limits, is on the Proctoring page.
Five kinds of signal, each timestamped against a single attempt: periodic camera
stills (not a video stream), screen images, short audio clips, fullscreen and
focus events, and device fingerprint snapshots. Every stream is capped per
evidence kind and event type, so a misbehaving client cannot flood the record.
No. Proctoring observes and records; it never refuses an attempt.
There is no risk score, no verdict, and no automatic rejection. Everything merges
into one time-ordered timeline that a person on your team reads and judges.
Nothing we capture can end an honest participant's exam.
Not in the absolute, and we will not pretend otherwise. A second laptop, a phone,
or another person in the room defeats any software running on the exam machine.
What the platform does is deter and collect evidence for review.
What is genuinely solved: exam traffic is protected by HTTPS and correct answers
are never sent to the exam device, so there is nothing on the wire worth
intercepting.
Exams are sat in the browser, which cannot enumerate running processes, detect a
virtual machine, or notice a USB or HDMI device being plugged in. So a quiet
timeline honestly means "we could never look", not "we looked and
found nothing" — an absence of signals is not a clean bill of health.
Participants declare whether their camera and mic work; the platform does not try
to detect it. A camera the browser cannot open is indistinguishable from one that
was covered, and refusing the exam over it would strand someone whose hardware
genuinely failed. "Never asked" and "told us it was broken" stay distinct in the
record, so a reviewer can read a gap correctly.
Evidence bytes — camera stills, screen images, audio clips — live in object
storage; only references and event rows go in the database. Evidence is visible to
reviewers in the organisation that owns the attempt, and to no one else. See
Security and the
Privacy Policy.
The organisation running the assessment is the data controller; ZiniApps is the
processor and acts on its instructions. That means participant requests — access,
correction, deletion — go to the organisation that issued the assessment, not to
us. Details in the Privacy Policy.
Support
Getting set up and running assessments
The whole workflow is walked through on the Product page.
About ten minutes if your questions are ready. Import your MCQ question bank as
JSON, set a duration and how many questions each person gets, then start inviting.
There is nothing to install and nothing to schedule with us first.
As structured MCQ JSON — question text, up to five choices, a section, and the
correct answer. You can review every question, choice, and correct answer before
going live. There is no LLM processing: questions must arrive in
the expected schema, and the platform does not interpret unstructured files.
Each person receives a unique secret key with an expiry, emailed
when you create the assessment. They start the exam with that key — there is no
participant account for your team to create or manage.
Yes. Resending issues a fresh key with a new expiry against the same participant
and quiz, and emails it out again. The expiry is set when you create the
assessment — hours or days, whatever suits the exercise.
The attempt can be resumed — answers already submitted are preserved and the
exam picks up where it left off. Note that the timer does not stop: it starts when
the participant starts, and the assessment ends automatically once the configured
duration elapses.
Yes — participants can return to earlier questions and change answers until the
time runs out. Answering a question moves them forward automatically, and they
cannot pause or stop the attempt midway. They can raise an issue from the exam UI
if something goes wrong.
Yes, as ADMIN or STANDARD users scoped to your organisation, at no extra cost —
there is no per-user charge. Every record belongs to your organisation and is
never visible to another.
Every attempt is reported section by section rather than as a single number. You
can search and filter results across participants, quizzes, and profiles, open a
detailed view for any attempt, and export any filtered result set to
CSV.
Use the forgot-password flow at sign-in. Resets use a one-time code sent to your
email rather than a click-through link, so there is no reset URL to leak or
replay. Administrators cannot change another user's password.
Create a ticket on our Contact page. One queue covers
sales, support, and security, so you never have to pick the right one — and you
get a reference you can track. We reply by email.