SimPrep AI is aimed at candidates who want AI-patient consultation practice, and it markets itself for both medical-student OSCEs and the MRCGP SCA with "RCGP-aligned feedback from senior GPs." It is a genuine AI-patient simulator, which is the archetype the SCA search intent expects. The principal limitation to know before you rely on it: it is a multi-exam tool rather than an SCA-exclusive one, and at last check its public pages exposed few verifiable specifics — case counts, price and whether patients are voice or text — so several figures here must be verified on the product page.
What SimPrep AI offers for MRCGP SCA right now
Checked against the vendor's public pages on 19 July 2026. SimPrep AI describes itself as AI-simulated medical consultation practice for "OSCE and SCA exam prep," with "instant RCGP-aligned feedback from senior GPs" and a free consultation to start. Beyond that, the public-facing material at last check did not confirm the details a candidate needs: the number of SCA cases or scenarios, the access period, the subscription price, and whether the AI patients are voice-based or text-based were not clearly stated — so treat all of these as to-be-verified rather than assumed, and confirm them on the product page before purchase. What can be said honestly is that it is positioned as a genuine AI-patient simulator with GP-referenced feedback, spanning two different examinations; everything more specific is vendor-reported and, at the time of writing, thinly evidenced in public, which is itself a finding worth weighing.
The exam your practice must answer to
The SCA is twelve simulated twelve-minute consultations, 144 minutes total, sat remotely from ST3, nine diets a year, fee around £1,207. Role-players consult by video or telephone through the RCGP platform; you take a history, request examination findings verbally, and state management aloud. Marking is by calibrated examiners across three domains — Data Gathering and Diagnosis, Clinical Management and Medical Complexity, and Relating to Others. This exam anchor matters especially for a multi-exam tool: a medical-student OSCE and a GP SCA differ in level, context and marking, so "RCGP-aligned" is a claim to check against the RCGP's own candidate guidance rather than to take on trust. The official pages are the authority; a vendor's alignment statement is a third-party claim.
Format map: which SCA tasks SimPrep AI reproduces
Because several specifics are unconfirmed, this format map is as much a checklist to verify as a description — which is the honest way to audit a thinly-documented tool.
| SCA task | Reproduced by SimPrep AI? | What to verify |
|---|---|---|
| Spoken consultation with an AI patient | Likely (AI patients) | Confirm whether patients are voice or text |
| 12-minute timed consultation | Verify | Confirm whether the SCA clock is enforced or self-imposed |
| History and data gathering | Yes | Central to any consultation simulator |
| Verbal management planning | Yes | Confirm it is marked and referenced to guidelines |
| Examination findings on request | Verify | Confirm how requested findings are surfaced |
| The three examiner domains | Verify | Confirm feedback maps to the three SCA domains, not generic OSCE marking |
| Case mix depth for SCA specifically | Verify | Confirm SCA case count and GP-context calibration, given the OSCE overlap |
| A full 12-case circuit | Verify | Confirm whether a full diet-length mock is available |
| Interpersonal rapport and non-verbal signals | No | No automarker judges genuine rapport |
The pattern is deliberate: for a well-documented simulator most of these rows resolve to a confident "yes" or "no," but here several resolve to "verify," and a candidate should read that as a prompt to trial the free consultation and check the specifics before committing money or reliance.
Fidelity test: timing, interface, scoring and permitted actions
On timing, confirm whether SimPrep AI enforces the twelve-minute clock or leaves it to you; pacing is part of what the SCA tests. On interface, no third-party tool mirrors the RCGP platform, so treat it as rehearsal. On scoring, the key fidelity question for a multi-exam tool is whether SCA feedback is genuinely mapped to the three SCA domains or to a generic consultation rubric shared with the OSCE product — verify this, because domain-referenced feedback is what makes SCA practice actionable. On permitted actions, confirm that management is kept verbal, matching the exam. On branching — a header theme of this audit — verify how the AI patient responds to your questions: a high-fidelity simulator lets the consultation genuinely branch on what you ask and miss, while a shallow one follows a fixed script regardless; trial the free consultation specifically to test whether the patient adapts to you or merely recites.
Case-mix audit: is the spread realistic?
The case-mix question is sharper for SimPrep AI than for an SCA-only tool, because its library is shared across OSCE and SCA use. Verify how many cases are genuinely SCA-level and GP-contextual rather than student-OSCE scenarios repackaged, since the SCA samples undifferentiated primary-care presentations, multi-morbidity, mental health, women's and sexual health, paediatrics, and ethically and safety-loaded consultations in GP context. A tool built first for OSCEs may under-represent exactly the GP-specific complexity the SCA rewards. Whatever the count, audit your practised distribution against the RCGP's published case content rather than the vendor's headline, applying the discipline in completion is not coverage: breadth is a property of your practice, not the library.
Feedback audit: observable, inferred and generated
SimPrep AI's "RCGP-aligned feedback from senior GPs" still resolves, in use, into three layers you must weight differently. Observable behaviours — red-flag screening, agenda-setting, safety-netting, sharing a plan — are checkable and reliable. Inferred competence — the model's read of your reasoning and management quality — is directional. Model-generated commentary is fluent but can mis-state you, and any interpersonal judgement is the least reliable of all because rapport is substantially non-verbal. The "from senior GPs" framing is worth probing: verify whether it means the feedback templates and exemplars were authored by GPs (a quality signal for the observable layer) or that GPs review your individual consultations (a stronger, and rarer, claim). Either way, calibrate by observability using the method in our pillar on calibrating automated feedback before you trust the score.
Repetition risk and preserving unseen cases
Until you have verified the SCA case count, assume the library is finite and treat repetition risk accordingly: memorising a limited set produces false fluency, where your second run through a remembered case feels smooth but rehearses recognition rather than the unseen-problem-solving the exam demands. If the count turns out small, the risk is higher and rotation is essential; if it is large or generated, the risk shifts to converging on the model's house style. Either way, preserve a reserve of cases you have never worked through for the final fortnight, so your last timed rehearsals measure cold performance. The safest late-stage calibration mixes preserved unseen cases with a separate, GP-written mock rather than replaying SimPrep AI's familiar material.
What it cannot test
SimPrep AI cannot reproduce examiner variability across a real diet, nor the RCGP's exact platform and remote logistics. As a multi-exam tool it cannot guarantee GP-level SCA depth without verification. No automarker, including this one, can judge genuine non-verbal rapport, so Relating to Others is only partly rehearsed and not reliably scored. And thin public documentation means you cannot fully assess its case mix or branching without trialling it yourself — which is a limitation of transparency as much as of capability. Name these honestly and you will use the free consultation to answer them rather than assuming the marketing settles them.
Where iatroX honestly sits
iatroX is not a consultation simulator and does not replace SimPrep AI's AI-patient practice — state that plainly. iatroX is the knowledge layer beneath the consultation: unseen multiple-choice measurement and current clinical knowledge. When SimPrep AI's feedback flags weak management, or when you cannot tell whether a case was GP-calibrated, iatroX is where you check the management against current UK guidance and test, on unseen questions, whether that knowledge holds. It measures the medicine, not the manner, and it does so on material neither you nor a simulator selected. Keep the caveat from why your Q-bank percentage is not your exam score in view: no single score, from a simulator or a bank, is a pass prediction.
A seven-day pattern for busy ST3 trainees
This plan uses SimPrep AI for one defined job — AI-patient consultation rehearsal — and iatroX for the separate job of measuring clinical knowledge on unseen items, with no claim to any proprietary algorithm. Monday: start with the free consultation and test whether the patient genuinely branches on your questions; run one full case, timed. Tuesday: two SCA-level cases, feedback reviewed against observable behaviours only. Wednesday: a knowledge session in iatroX's MRCGP SCA area on the management points flagged weak. Thursday: two cases across GP-specific groups you suspect the OSCE-leaning library under-supplies. Friday: a peer-observed role-play so a human judges rapport. Saturday: a timed mini-circuit of three cases, reviewed by domain. Sunday: rest, preserving unseen cases. SimPrep AI rehearses the consultation; iatroX keeps the medicine inside it current.
Continue, supplement, switch or stop
Continue if the free trial confirms genuine branching, domain-referenced SCA feedback and adequate GP-level case depth, and your management currency is improving. Supplement with human-observed practice for the rapport the automarker cannot judge, and with a GP-written mock for a full-fidelity circuit. Switch to an SCA-dedicated simulator if verification shows the SCA case mix is thin or the feedback is generic OSCE marking rather than the three SCA domains. Stop relying on unconfirmed vendor claims — resolve every "verify" in this audit before you commit your revision to the tool. Decide by measured gaps, not by novelty or sunk cost.
Three mistakes this audit is designed to stop
First, taking "RCGP-aligned" and "from senior GPs" on trust: verify whether feedback maps to the three SCA domains and what the GP involvement actually is, because a multi-exam tool can carry generic marking. Second, assuming OSCE case depth equals SCA depth — the SCA rewards GP-specific, undifferentiated complexity a student-OSCE library may under-represent. Third, committing money or revision time before resolving the unconfirmed specifics; the honest response to thin documentation is to trial the free consultation and check the product page, not to assume the gaps are filled.
Frequently asked questions
Is SimPrep AI enough for MRCGP SCA on its own? Possibly for the consultation-rehearsal half, if verification confirms genuine branching, domain-referenced SCA feedback and adequate GP-level case depth — but it cannot judge rapport reliably or measure your clinical knowledge on unseen questions, and its multi-exam scope means you should confirm SCA specifics before relying on it, so pair it with human-observed practice and an unseen knowledge check.
Which MRCGP SCA component does SimPrep AI not reproduce well? The Relating to Others domain and, potentially, GP-specific case depth: no automarker judges genuine non-verbal rapport, and a tool shared with OSCE preparation may under-represent the undifferentiated primary-care complexity the SCA rewards — verify the SCA case mix before you rely on it.
How many unseen SimPrep AI cases or stations should I preserve for final MRCGP SCA calibration? Preserve at least twelve unseen cases for the final fortnight, and because the case count is unconfirmed, plan to calibrate on a separate GP-written mock as well, so your last timed run does not depend on a library whose size and depth you have not verified.
When should I stop using SimPrep AI and move to mixed mocks? When your consultation structure, timing and management currency are stable — usually the final two to three weeks — move to full timed twelve-case circuits, ideally GP-written and human-observed, rather than continuing single-case AI practice.
How should I combine SimPrep AI with iatroX without duplicating practice? Use SimPrep AI to rehearse the consultation and iatroX to keep the clinical management inside it current on unseen questions — the simulator flags a thin plan, iatroX tells you what the plan should have been. This is the two-source design in the two-Q-bank rule, applied across a simulator and a knowledge bank; our comparison hub helps you place both.
The bottom line for ST3 trainees
The honest verdict on SimPrep AI for the SCA: a genuine AI-patient simulator with GP-referenced feedback, but a multi-exam one whose public specifics were thin at last check — so its value depends on what the free consultation and product page reveal about case count, price, voice-versus-text, branching depth and whether feedback truly maps to the three SCA domains. Trial it deliberately to resolve those questions; do not commit revision or money on the strength of the marketing alone. If verification is favourable, use it for the consultation-rehearsal job, add human observation for rapport and a GP-written mock for a full circuit, calibrate the automated feedback by observability, and let an unseen knowledge source measure whether your management is current. Used that way, SimPrep AI rehearses the consultation, and iatroX keeps the medicine inside it honest.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; SimPrep AI's positioning (AI-patient practice for OSCE and SCA with RCGP-aligned feedback and a free consultation) is vendor-reported, and at that date its public pages did not confirm the SCA case count, access period, price, or whether patients are voice or text — verify each on the product page. MRCGP SCA format and its three marking domains are per the RCGP; "RCGP-aligned" is a vendor claim to check against RCGP guidance. Disclosure: iatroX operates a competing clinical-knowledge and question-bank platform and is not a consultation simulator, so this audit confines iatroX's role to the unseen knowledge measurement the product does not claim. Corrections via the feedback route on iatrox.com. References: RCGP Simulated Consultation Assessment pages (rcgp.org.uk); SimPrep AI product pages (simprep.ai); related reading: calibrating automated feedback before you trust the score and why your Q-bank percentage is not your exam score.