Candy and Frank is a personal AI study companion for Australian students in Years 3–8. It finds the skills your child has not mastered yet, teaches those until they hold, and plans the weeks to the exam — so the time at the desk turns into knowledge that stays.
No child is average. So we never teach to one.
Less time at the desk.More of it understood. A 40-minute diagnostic, then eight questions a session on the one skill that is missing.
One subject. One child. One plan.Nothing generic, nothing improvised. Ask it about anything else and it has nothing to say.
Mastery is what counts.Every student who works through this tutor properly, week by week, masters the skills their exam needs — and you watch it happen, question by question. That is our promise.
The diagnostic samples every band of the real paper. From then on each weekly paper targets the skill that is actually missing — never a page of sums your child can already do.
Each miss is traced to the step it went wrong at, explained step by step with a sketch, then checked with a fresh question of the same skill, solved unaided.
From the results, the target, the weeks left and your sessions per week: one skill each week, a checkpoint every four, and a plain time check when a target does not fit.
Pick the school year; the diagnostic is built for that year's paper.
The Australian Mathematics Competition, sat in the first week of August. 16 of the 17 measured topics, three difficulty bands like Q1–10 / Q11–20 / Q21–30, figure questions, A–E and 0–999 answers like the real paper.
Start the free diagnosticThe same engine tuned to the Upper Primary paper, where the award is decided in Q21–30: verified anchors for the hard tail, weekly papers on the skill that is actually missing, a 30-question mock on the clock.
Start the free diagnosticSat in April or May of Year 4, by invitation. The school publishes no paper, so this track trains what every public Year 3–5 assessment tests at that age: numeracy, spelling, grammar, punctuation and thinking skills. Reading joins soon; no writing task is guessed at.
Start the free MGS diagnosticMathematics, English, Science and the Spelling Bee — the Medal-deciding final third of each paper.
Numeracy and language conventions, built from the official item-by-item descriptors.
The Junior paper for the first years of secondary school.
Want to be told first when one of these opens? Email wcandy@candyandfrank.com with the year and the test.
Everything below was produced for a sample student. Nothing was written by hand.
The diagram shows an L-shaped garden. Only two sides are labelled. What is the perimeter of the garden, in centimetres?
Warm-up like questions 1–10 of the real paper
Open the sample testanswered correctly, by band — with each miss explained and the skills to master next.
Open the sample resultsFrom the results, the target award, the weeks left and the sessions per week: which skill each week, a checkpoint test every four weeks, two rehearsal weeks before the exam.
See a sample planPer child, per program. Monthly stops when you stop; the term does not auto-renew. Prices are shown before anything is charged.
See membershipMeasured from 300 official Middle Primary questions, not guessed. The test samples the topics we can generate and verify; the rest are named honestly as not yet covered.
Covered means the engine can generate and verify questions on that topic. It says nothing about your child yet: a 12-question diagnostic samples the bands, it does not settle 16 topics — the weekly papers and checkpoints do that, one skill at a time.
Yes. Years 3–4 sit AMC Middle Primary and Years 5–6 Upper Primary, and the test is built for that division: every question comes from a family modelled on a real question of that paper (81 Middle Primary and 115 Upper Primary anchors, each proven against the official answer). The test ramps the way the real paper does: warm-ups like Q1–10, then Q11–20, then the award-deciding Q21–30. The Melbourne Grammar track is for Year 4, when the school tests.
A general chatbot is not looking things up; it is predicting the next word, and on a competition question it can be confidently wrong. This tutor does one job. Every question is generated and verified by code before your child sees it, marking is code comparing numbers, and the explanations were built offline and checked. Ask it about anything else and it has nothing to say — that is the point.
The question families, methods and explanations were built with AI assistance, offline, then checked by code. When your child sits a test nothing is improvised: every question was generated and verified by code, and marking is code comparing numbers. That is why every result can cite its evidence.
A class is taught to the average of the room, so a strong child waits and a struggling one is rushed. Here the diagnostic samples every band in 40 minutes, then each weekly paper is eight questions on the one skill that is missing, with the miss explained at the step it broke. Nothing your child already does right is drilled again.
Yes — on a tablet or laptop, with a friendly timer, or you can print it and type the answers in afterwards. The real competition allows a calculator in Middle and Upper Primary, and so do we; the questions are the reasoning kind where it does not help much.
The diagnostic test, its results and the explanations are free. The personal study plan and the weekly papers are membership: $39 a month, or $99 for a 12-week term that does not auto-renew. Per child, per program. Prices are shown before anything is charged.
That depends on how seriously your child uses it. We are confident that a student who works through the weekly papers, the explanations and the fresh questions makes visible progress — and you see that progress yourself, question by question, on every results page. We plan in skills, and we say plainly when a target does not fit the time left, then offer the nearest one that does.
Melbourne Grammar entry is a Year 4 test; the registration page checks the pair for you. No card, no lock-in.