This workflow is for internationally trained dentists preparing for the NDEB Assessment of Fundamental Knowledge (AFK) who use QuizOdontist and its AI tutor, and who want the tutor to sharpen their reasoning rather than short-circuit it. It addresses the AFK's single knowledge component. The principal limitation is one every AI-tutor user must hold onto: a language model can explain fluently and still be wrong or out of jurisdiction, so its output is a hypothesis to verify, not an authority to trust.
Used well, an AI tutor is a Socratic partner that makes you defend an answer before it comments; used badly, it is an answer-vending machine that produces the feeling of understanding without the substance. The twenty-five prompts below are built to force the first mode.
What QuizOdontist offers for NDEB AFK right now
Figures are vendor-reported and last checked on 20 July 2026; confirm on quizodontist.com, because the AFK-specific figures matter more than the headline totals.
| Item | What we found (last checked 20 July 2026) |
|---|---|
| Product type | Web-based dental board question bank and lessons with an AI tutor; covers AFK (Canada), INBDE and ADAT (US) |
| Question volume | "50,000+ questions" and "2,800+ lessons" across all three exams — vendor-reported; this is not an AFK-only figure, so verify the AFK-specific count |
| AI / adaptive features | A "24/7 AI Tutor" that "reads your performance" and "suggests what to fix next"; Premium includes 30 AI Tutor credits per month; performance tracking — vendor-reported |
| Price | Free tier (5 lessons per subject); Premium $399 CAD for 3 months (about $133/month); an AI-credit add-on at $20 CAD one-time — vendor-reported |
| Access | Premium in 1, 3 or 6-month terms; free tier ongoing — vendor-reported |
| AFK coverage | AFK is a 200-MCQ knowledge exam, and QuizOdontist's MCQ-plus-explanation format maps to it; it does not cover the later clinical stages of the NDEB Equivalency Process |
Two flags to carry forward. First, the platform serves three exams, two of them US (INBDE, ADAT), so its content and the tutor's default sources may lean US-calibrated while the AFK is Canadian — localisation must be actively checked. Second, the "50,000+ questions" total spans all three exams, so do not assume it all applies to the AFK.
The NDEB AFK exam: what you are preparing for
The AFK is the first examination in the NDEB Equivalency Process for internationally trained dentists. It consists of 200 single-answer multiple-choice questions delivered in two parts, two hours per part, four hours in total. It is offered by electronic delivery at Prometric test centres and in booklet format at select sites. Each question is worth one point, and a test-equated, re-scaled score of 75 or greater is reported as a pass. It tests biomedical science and applied clinical science, and you may attempt it a maximum of three times. A pass is required before you continue to the later examinations in the Equivalency Process (confirm the current protocol on ndeb-bned.ca).
Two consequences for AI-tutor use follow. Because the AFK is single-best-answer, the discipline that pays is precise option elimination, not essay-style breadth. And because it is Canadian, every management or standard-of-care claim the tutor makes should be checked against a primary or Canadian source — the SmPC or the product monograph for a drug fact, a current Canadian guideline for a management threshold — and never accepted just because it sounds authoritative.
Six prompt types that test reasoning, not recall
Good tutor prompts fall into six families, each targeting a different failure: mechanism (do I understand why?), discriminating features (can I tell look-alikes apart?), option elimination (can I kill the distractors?), guideline verification (is this the current standard, and whose?), jurisdiction and localisation (does this hold in Canada?) and counterfactual and retrieval testing (can I still do it next week, and when the variables change?). The twenty-five prompts below are grouped by these families and by the error type each is built to catch.
25 reusable prompts grouped by error type
Keep these generic — never paste proprietary question text into the tutor. Paste your own reasoning and let the prompt do the work.
Error type 1 — Shaky basic science (mechanism prompts).
- "Explain the mechanism behind the correct answer at the level of cell biology or pharmacology, then tell me which single step, if I misremember it, would flip me to the most tempting wrong option."
- "Do not give me the answer yet. State the physiological or microbiological principle this item is testing, and I will predict the answer from it."
- "Reduce this item to the one basic-science fact it depends on; if I can state that fact, what else does it let me answer?"
- "Name the enzyme, receptor or pathway involved and one clinical consequence if it fails — nothing more."
Error type 2 — Confusing look-alikes (discriminating-feature prompts). 5. "List the two or three conditions most often confused with the correct answer here, and give me the single discriminating feature for each." 6. "Give me a two-column comparison: the feature present in the correct diagnosis versus the feature present in its nearest mimic. No commentary." 7. "If one detail in the history or radiograph changed, which competing diagnosis would become correct, and what is that detail?" 8. "Which clinical or radiographic finding, if absent, would rule the correct answer out?"
Error type 3 — Falling for distractors (option-elimination prompts). 9. "Take each of the five options and give me a one-line reason it is right or wrong — no summary paragraph." 10. "Which distractor is designed to catch someone who half-knows this topic, and what half-knowledge does it exploit?" 11. "Rank the options from most to least defensible, and tell me what would have to be true for the second-ranked option to win." 12. "Which two options can be eliminated before any calculation or recall, and on what principle?"
Error type 4 — Outdated or misremembered standards (guideline-verification prompts). 13. "State the guideline or standard of care this item rests on, name the issuing body, and give the year of the version you are quoting." 14. "Is the management here a threshold-based decision? If so, give me the exact threshold and its source." 15. "Give me the regimen — agent, dose, timing, route — and flag any part that is contested or has changed recently." 16. "What is the current first-line, and why was the previous first-line superseded, with dates?"
Error type 5 — Applying US practice to a Canadian exam (jurisdiction and localisation prompts). 17. "Does this answer reflect Canadian and NDEB conventions specifically, or US and INBDE practice? If they differ, give both and tell me which the AFK expects." 18. "Is any drug name, material or protocol in this answer country-specific? Give the Canadian term and note the US equivalent." 19. "Which NDEB blueprint domain does this item map to, and would a Canadian regulator phrase the standard differently?" 20. "Flag any part of your explanation you are inferring from US sources rather than Canadian ones, and rate your confidence."
Error type 6 — Knowing it today, forgetting it next week (counterfactual and retrieval prompts). 21. "Do not explain yet. Ask me one question that would prove whether I actually understand this, then wait for my answer." 22. "Change one variable in the scenario so the answer changes, and ask me the new answer." 23. "Give me a transfer question on the same principle set in a different clinical scenario, and mark my response." 24. "Summarise the single rule I should carry from this item in one sentence, then test me on it with a fresh scenario." 25. "Note this concept for a spaced re-test; when I give you a date one week out, re-ask it as a new question without showing the original."
The anti-answer-leak rule
Before you open the tutor on any item, commit — out loud or in writing — to an answer and a one-line rationale. Only then do you engage the AI, and your first message should be your reasoning, not "what is the answer?". This single rule separates a tutor that builds competence from one that launders it. Let the model speak first and you will nod along to a fluent explanation and learn almost nothing; commit first and every reply either confirms a genuine inference or exposes a real gap. The credits you spend then buy diagnosis, not answers.
The jurisdiction check
Because QuizOdontist serves US exams too, make prompt 17 a fixed habit. For any answer touching management, materials, medication or standards of care, ask the tutor to state explicitly whether it is giving Canadian and NDEB-aligned guidance or US practice, and to name the date of the supporting guidance. Antibiotic-prophylaxis regimens, radiographic-selection criteria, referral pathways and drug availability are where US and Canadian answers quietly diverge, and the AFK expects the Canadian one. When the tutor cannot say which jurisdiction it speaks for, treat the answer as unverified and check a primary Canadian source.
Build a misconception record
Every confirmed error earns a four-line entry, and no more: one sentence stating the incorrect rule you were using, one stating the corrected rule, one transfer question that tests the corrected rule in a new setting, and one review date. It is deliberately short so you will actually keep it. The value is in the first two lines: naming the wrong rule explicitly is what stops it recurring, because a misconception you can articulate is one you can catch mid-exam. The transfer question and review date turn the correction into spaced retrieval rather than a note you never reread.
Run a weekly verification sample
Once a week, take a small random sample — five to ten — of the tutor's explanations from that week and independently check them against official or primary sources: the NDEB blueprint and sample questions, current Canadian guidelines, and the SmPC or product monograph for any drug claim. Log every discrepancy you find. This does two things: it catches specific errors before they calcify, and it builds a personal sense of where this tutor is reliable and where it drifts — more useful than a generic warning that AI can hallucinate. When the same category of error keeps recurring, say US-default prophylaxis regimens, you know exactly which prompt to deploy harder.
A seven-day plan for an internationally trained dentist
One week balancing content review with local-convention checking, using QuizOdontist for one job — practice items and AI-mediated reasoning — and a bank such as iatroX for unseen transfer practice. No proprietary-algorithm claims: "adaptive" means you route your own errors back into the plan.
- Monday: 40 minutes. A timed AFK-style set; for each miss, commit to a rationale, then run a mechanism or discrimination prompt.
- Tuesday: 30 minutes. Convert Monday's misses into misconception-record entries; run jurisdiction checks on any management items.
- Wednesday: 40 minutes. New topic set; option-elimination prompts on the near-misses; log gaps.
- Thursday: 30 minutes. Unseen transfer questions on Monday–Wednesday topics in a separate bank; no repeats.
- Friday: 30 minutes. Guideline-verification prompts on this week's management items; update any dated standards.
- Saturday: 45 minutes. Weekly verification sample against primary sources; log discrepancies. Interleaved mixed block, timed.
- Sunday: 20 minutes. Review misconception records due today; set next review dates.
Decision checklist: continue, supplement, switch or stop
- Continue with QuizOdontist and its tutor while your verification sample stays clean and your unseen accuracy is rising.
- Supplement with unseen questions from another bank as soon as you are re-seeing items or your practice has become recognition rather than retrieval.
- Switch your emphasis to primary sources for any domain where the weekly sample repeatedly catches the tutor in jurisdiction or currency errors — trust the blueprint and the guideline over the model there.
- Stop grinding a topic when timed, unseen accuracy has plateaued and errors are careless; move to weaker domains.
Decide on measured signals — verification-sample discrepancy rate, unseen-score trend, error type, weeks to exam — not on novelty or the credits you have already bought.
Frequently asked questions
Is QuizOdontist enough for NDEB AFK on its own? It can be a strong core, because the AFK is entirely single-best-answer MCQ and QuizOdontist's format and AI tutor map onto that directly. But any single source becomes recognition once you have cycled its items, and an unchecked AI tutor can drift towards US practice or dated standards. Supported by a second unseen bank and a weekly primary-source verification habit, it is enough; treated as a closed loop you never check, it is not.
Which NDEB AFK component does QuizOdontist not reproduce well? The AFK is a single written MCQ exam, so there is no oral or OSCE component to miss at this stage, and QuizOdontist's format matches it well. The honest gap is elsewhere: it does not reproduce the later clinical stages of the Equivalency Process, such as the clinical judgement and clinical skills assessments, and its AI tutor does not reproduce a verified primary source. iatroX shares that second boundary — it is a knowledge and unseen-MCQ layer, not a clinical-skills simulator and not an authority to trust without verification.
How should I verify QuizOdontist AI answers for NDEB AFK? Commit to your own answer first, then require the tutor to name its sources and their dates, and run the jurisdiction prompt on anything involving management, medication or materials. Once a week, independently check a random sample of its explanations against the NDEB blueprint and sample questions, current Canadian guidelines and the SmPC or product monograph for drug facts, and log discrepancies. The goal is to learn where this tutor is reliable for you and where it needs a primary-source backstop.
When should I stop using QuizOdontist and move to mixed mocks? Move to full-length, timed, mixed mocks once your accuracy on unseen items is stable and your weekly interleaved blocks stop improving — usually the final few weeks. The marginal value of more single-topic practice then falls, and what you need is exam-condition rehearsal across the whole blueprint under the four-hour, two-part structure, which surfaces the pacing and stamina issues that topic drilling hides.
How should I combine QuizOdontist with iatroX without duplicating practice? Give them separate roles and never cross them: QuizOdontist for AI-mediated reasoning on its items, iatroX for fresh unseen measurement and the weekly mixed block. The two-Q-bank rule linked below is the safeguard — do not answer the same item in both, log each error once, and keep one tool for learning and one for measuring. Your accuracy then reflects transfer to new questions rather than memory of ones you have seen, which is the figure that actually predicts the exam.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. Figures attributed to QuizOdontist are vendor-reported and were current at the last-checked date; the headline question and lesson totals span AFK, INBDE and ADAT, so verify the AFK-specific figures, features and prices on quizodontist.com before relying on them. Disclosure: iatroX operates a competing question bank and clinical-knowledge platform; this article confines iatroX's role to the unseen-measurement and knowledge jobs an AI-tutor product does not verify for you, and iatroX makes no claim to reproduce the clinical stages of the NDEB Equivalency Process. Corrections are welcome via the feedback route on iatrox.com.
References: National Dental Examining Board of Canada — Assessment of Fundamental Knowledge, format, scoring and Equivalency Process (ndeb-bned.ca); QuizOdontist product pages (quizodontist.com, vendor-reported); iatroX Canada exam hub (iatrox.com/canada); "Your Q-Bank Percentage Is Not Your Exam Score"; "How to audit an AI medical-exam tutor" and the two-Q-bank rule framework pillars; the iatroX comparison hub.