SampChamp is a CCFP preparation platform whose headline feature is "instant" AI feedback on SAMP answers, plus an "AI patient" mode for oral practice in its higher tier. This audit is for candidates deciding whether to rely on that feedback: it sets out what an automated marker does well, where it can quietly drift, and a verification protocol you run yourself. The honest headline first — SampChamp's "AI tutor" is essentially an automated marking-and-feedback engine, not a Socratic conversational tutor, so you audit it as a grader whose scheme could be outdated, not as a reasoning partner.
What SampChamp offers for CCFP right now
Figures below are vendor-reported from sampchamp.ca and were last checked on 19 July 2026. Confirm current numbers and pricing on the product page before you buy.
| Item | Detail (vendor-reported, last checked 19 July 2026) |
|---|---|
| Exam covered | CCFP — SAMP practice plus SOO "AI patient" preparation (Pro tier) |
| Volume | 300+ cases across all 105 priority topics; 20-case and 40-case timed exams |
| "AI tutor" | Instant automated marking with a breakdown of what you missed and why; methodology not disclosed |
| Extras | Performance analytics dashboard; Anki flashcard generation |
| Access / price | Basic $297 one-time; Pro $447 (listed as reduced from $744); access stated to run to the 2026 exam day; 5 free tutor cases; 7-day money-back guarantee |
| Not reproduced | The live SOO performance; the MCQ and short-menu format entering the written exam from 2026 |
Two things to hold in mind. First, the pricing and the access window are tied to the current exam cycle and are vendor-reported as of the check date — confirm them. Second, "methodology not disclosed" is itself an audit finding: when a vendor does not say how its AI marks or what sources it grounds against, the burden shifts to you to verify.
The CCFP exam you are actually preparing for
The CFPC Certification Examination in Family Medicine has a computer-based SAMP written component of four hours and an oral component of five 15-minute Simulated Office Orals on a separate day. The College does not publish a fixed per-component question count; verify current figures on cfpc.ca. From April 2026 the SAMP begins moving from written free-text toward multiple-choice and short-menu formats — up to 25 per cent of Fall 2026 cases, all cases from 2027 — so confirm the format for your sitting. A free-text marker trains the reasoning the exam rewards, but the answer format it drills is being phased down; weight your audit toward the durable question, which is whether the clinical feedback is correct and current.
Testing methodology: a fixed rubric you can reuse
Do not judge an AI marker on the case that flatters it. Run a fixed, representative set through the tutor and score every one on the same rubric. Use six item types: a straight-recall item; a diagnosis item; a next-investigation item; a management item; an ethics or professional item; and one deliberately ambiguous item where reasonable clinicians differ. For each, record whether the feedback was correct, whether it was grounded in a named current Canadian source, whether it handled uncertainty honestly, and whether it added anything beyond the marking scheme. The same rubric, every time — that is what makes it an audit rather than an impression, and it mirrors the discipline in the AI-tutor audit pillar.
Grounding audit: can you see where the answer comes from?
The most important question about any AI feedback is provenance. When SampChamp tells you what you missed, ask: is this citing a specific Canadian guideline (Hypertension Canada, Diabetes Canada, the CTFPHC, the SOGC, the CFPC's own guidance), an external source, or uncited model output? You can inspect this without reproducing any copyrighted content: note whether the feedback names a source and year, and cross-check that named source against the primary document. Feedback that asserts a threshold or first-line agent with no attribution is the highest-risk category — not necessarily wrong, but unverifiable, and unverifiable guidance is not something to memorise on a high-stakes exam.
Reasoning behaviour: grade it for what it is
Because SampChamp marks rather than converses, several "tutor" behaviours simply do not apply, and saying so is part of an honest audit. An automated marker does not ask you useful diagnostic questions, does not build a differential with you, and cannot be argued with when it is wrong. It also, by design, reveals the marking scheme — so "premature answer reveal" is structural, which is exactly why the commit-before-reveal discipline has to come from you. Where SampChamp does have a conversational surface — the SOO "AI patient" mode — audit that separately: does it stay in role, does it volunteer answers a real patient would not, and does it correct or reinforce a false premise you introduce? Treat any claim about that mode as vendor-reported until you have tested it.
Exam fidelity: does the advice fit CCFP?
Fidelity is where generic AI most often slips. Test whether the feedback respects CCFP-specific jurisdiction (Canadian, not US or UK guidance), the College's terminology and priority-topic framing, the four-hour time constraint, and the blueprint's weighting. A marker that scores your management of hypertension against an American target, or that frames a case in non-Canadian screening terms, is producing plausible feedback that would cost you defensible marks. Localisation error is the single most consequential failure mode for a Canadian family-medicine tool, so weight it heavily.
Failure modes to watch for
- Hallucinated citations: a named guideline or figure that does not exist or does not say what is claimed.
- Overconfident wording: definitive language on a point where guidance is genuinely uncertain or contested.
- Outdated guidance: a superseded threshold or first-line agent, because a marking scheme is only as current as its last update.
- Answer leakage: structural in a marker, so enforce commit-before-reveal yourself.
- Plausible-but-unexamined elaboration: fluent extra detail that sounds authoritative but was never checked against a source.
These are illustrative categories to watch for when you run the rubric, not claims that SampChamp exhibits any of them; the point of the audit is that you find out for your own sitting.
Safe-use protocol: answer first, interrogate second, verify third
- Answer first. Write your full free-text answer and a one-line rationale, and submit, before you read any feedback. This preserves active recall and stops the marking scheme from doing your thinking.
- Interrogate second. Read the feedback and ask the fixed questions: what is the source, what is the year, what discriminates this diagnosis, why is each distractor wrong? For an ambiguous item, ask whether the feedback acknowledges the uncertainty.
- Verify third. For any point you intend to memorise, confirm it against the primary Canadian source. High-value prompts: "Which Canadian body sets this, and in what year?"; "Where does this differ from US or UK guidance?"; "What single finding would change this answer?" Log any discrepancy, exactly as you would when calibrating any AI-graded feedback before trusting the score.
Worked example: a seven-day week around clinical work
Give SampChamp one job — AI-marked SAMP feedback under audit — and iatroX one job — transfer practice on fresh, unseen items. No proprietary-algorithm claims are needed; a second non-overlapping bank simply supplies clean stems.
- Monday: 6–8 SampChamp cases, committed before reveal. Run the three-step protocol on each miss.
- Tuesday: for principles you missed, answer fresh iatroX items on the same principles; right cold means retire, wrong means step back.
- Wednesday: 6–8 new cases; run the grounding audit — is every substantive point sourced and current?
- Thursday: your weekly verification sample — five items checked against primary Canadian sources; log discrepancies.
- Friday: spaced retrieval of last week's corrected rules, from memory.
- Saturday: a 20-case timed SampChamp exam; read pacing and unseen first-attempt accuracy.
- Sunday: rest or an error-log review; pick next week's weakest topic and, if using it, a short SOO "AI patient" session.
Decision checklist: continue, supplement, switch or stop
- Continue if SampChamp's feedback survives your audit — grounded, current, CCFP-localised — and is still teaching you.
- Supplement with iatroX for unseen transfer once misses become recognition rather than knowledge gaps.
- Switch if your audit repeatedly finds ungrounded, outdated or non-Canadian feedback on high-weight topics and another resource is demonstrably better — a measurable finding, not a hunch.
- Stop new questions when coverage, unseen timed performance, pacing, retention and official-material calibration are all met, and move to official material and spoken SOO rehearsal.
Bottom line
SampChamp's instant AI marking can be a strong, time-saving feedback layer for CCFP SAMP practice, and its coverage of the priority topics is worth including in a revision stack. But it is a marker whose methodology is undisclosed, not a reasoning tutor, so the guardrails are yours to hold: commit before you reveal, audit provenance, weight Canadian localisation heavily, and verify anything you plan to memorise against the primary source. Treated as an audited feedback engine rather than an oracle, it earns its place; treated as an oracle, it is a route to memorising confident, unverified, possibly non-Canadian rules.
Frequently asked questions
Is SampChamp enough for CCFP on its own? For SAMP reasoning and priority-topic coverage it is a substantial resource, but "enough on its own" is the wrong standard. Its AI marking needs your verification to be safe, it does not reproduce the live SOO, and the written format it drills is changing from 2026. Pair it with official CFPC material for calibration, spoken practice for the oral, and a second bank for unseen transfer, and it becomes a strong component rather than a complete preparation.
Which CCFP component does SampChamp not reproduce well? The live Simulated Office Oral. SampChamp offers an "AI patient" mode in its Pro tier, which may help rehearse structure, but a text or scripted simulation is not the live, timed, verbal assessment with a standardised patient that the SOO actually is — and the vendor's description of that mode should be treated as vendor-reported until you test it. Its core strength is written SAMP feedback; the oral needs spoken rehearsal with a partner or a recognised course.
How should I verify SampChamp AI answers for CCFP? Use the three-step protocol and a weekly sample. Commit to your own answer before revealing the feedback; interrogate the feedback for its source and year; and verify any point you intend to memorise against the primary Canadian guideline. Once a week, take five random items and cross-check the marking points against Hypertension Canada, Diabetes Canada, the CTFPHC, the SOGC or the CFPC's own guidance, logging any threshold or recommendation that has moved. Undisclosed methodology makes this verification non-optional.
When should I stop using SampChamp and move to mixed mocks? Move to full-length, timed mocks per topic as each clears its first-attempt accuracy floor on unseen items, rather than when you finish the cases. Stop adding new SampChamp cases when coverage is complete, unseen timed performance is stable, pacing is comfortable and your weekly verification is turning up nothing new — at which point official material and spoken SOO rehearsal give a better return than more marked cases.
How should I combine SampChamp with iatroX without duplicating practice? Assign non-overlapping jobs. SampChamp is your AI-marked SAMP-feedback bank; iatroX is your unseen-measurement and transfer layer. When SampChamp marks you down on a principle, do not re-do that case — commit to a rationale on a fresh iatroX item testing the same principle, using the Socratic prompt before you reveal the explanation. That keeps your measurement uncontaminated by SampChamp's wording and turns two resources into an audit-then-transfer loop. The comparison hub sets out where each fits.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; all platform figures, including pricing and the stated access window, are vendor-reported and tied to the current exam cycle — verify them on the product page. SampChamp's AI methodology is not disclosed by the vendor; this audit treats it accordingly and does not assert any specific failure occurred, only the categories to test for. Disclosure: iatroX operates a CCFP question bank and a Socratic Tutor that compete with SampChamp; iatroX's role here is confined to unseen transfer practice and verification, jobs SampChamp does not claim, and it is not presented as an SOO simulator. Corrections via the feedback route on iatrox.com.
References: CFPC — Certification Examination in Family Medicine (cfpc.ca); SampChamp product and pricing pages (sampchamp.ca); iatroX CCFP bank (https://www.iatrox.com/canada/exam/ca-ccfp); "How to audit an AI medical-exam tutor" (https://www.iatrox.com/blog/how-to-audit-an-ai-medical-exam-tutor-grounding-answer-leakage-hallucinations-and-retention); "Your Q-Bank Percentage Is Not Your Exam Score" (https://www.iatrox.com/blog/qbank-percentage-not-your-exam-score).