Geeky Medics' SCA AI patients do something no question bank can: they let a GP trainee rehearse the timed, simulated consultations the MRCGP SCA is built around. This is a simulator audit, not an MCQ review — the question is what the tool faithfully trains across the exam's three domains, and where AI automarking only approximates a real assessor. It is for ST3 trainees deciding where an AI consultation simulator fits. The principal limitation: a model can score what you say, but it cannot fully judge clinical management complexity or the interpersonal nuance the SCA examiners weigh, so its feedback is a rehearsal aid, not a verdict.
What Geeky Medics offers for the SCA right now
Geeky Medics' SCA case platform provides 200 exam-style consultation scenarios with AI virtual patients you can type to or speak with, interactive mark schemes mapped to the RCGP domains (Data Gathering & Diagnosis, Clinical Management & Medical Complexity, Relating to Others), a Global Impression rating, an examiner walkthrough by the case authors, and performance analytics with domain-based scoring and timing. It is marketed around £10.99 per month; verify the current case count and price on the product page. The scenarios span RCGP Clinical Experience Groups, which is a real strength for case-mix breadth.
The exam that sets the bar
The MRCGP Simulated Consultation Assessment is 12 simulated consultations of 12 minutes each (144 minutes total), delivered remotely from a local GP surgery, taken during ST3, across nine diets a year, with a current fee of £1,207. It assesses three domains — Data Gathering and Diagnosis, Clinical Management and Medical Complexity, and Relating to Others (interpersonal skills) — testing whether a trainee can integrate clinical, professional and communication skills within a realistic time-limited consultation. Two features matter for a simulator: the strict 12-minute clock, and the fact that scoring rests on holistic examiner judgement across the three domains, not a checklist.
Format map: what the simulator reproduces
| SCA task | Simulator reproduces it? | Notes |
|---|---|---|
| 12-minute timed consultation | Yes, if you impose the clock | Core rehearsal value; untimed practice under-trains pacing |
| Data gathering through dialogue | Yes | Voice or text; strong for structured history |
| Clinical management within the consult | Partly | You can state a plan; the model's judgement of its quality is directional |
| Relating to others / interpersonal skills | Partly | Verbal empathy yes; non-verbal rapport no |
| Case-mix across Clinical Experience Groups | Yes | Scenario library spans presentations |
| Holistic examiner judgement | No | Automarking approximates domains; it is not a trained assessor |
| Remote-consultation logistics | Partly | Rehearses the format, not your exam-day setup |
The pattern: the simulator is strong on structure, pacing rehearsal and data-gathering dialogue, and only approximate on the management-quality and interpersonal judgement the real domains weigh.
Fidelity test
Compare the simulator's timing, interface and scoring categories with current RCGP SCA guidance. Two checks matter most. Timing: rehearse strictly to 12 minutes, because the SCA's difficulty is as much about completing a safe consultation in the time as about knowing the medicine — untimed practice builds a false fluency the clock will expose. Scoring: the mark schemes mapped to the three domains are a genuine aid for structuring practice, but the AI's domain scores and Global Impression are a model's inference, and the SCA is examiner-judged, so treat them as directional feedback, not a predicted result.
Case-mix audit
Check that the 200-case library spans common and rare, acute and chronic, and communication, ethics and safety scenarios in realistic proportions across the Clinical Experience Groups — rather than clustering on dramatic cases. Rotate deliberately across the groups the RCGP samples, including the mundane presentations that are easy to skip because they feel less like practice.
Feedback audit: observable, inferred, or generated?
The essential discipline with any AI examiner is separating three things its feedback blends: observable behaviours (did you screen for red flags, did you safety-net — reliable), inferred competence (its guess at your clinical reasoning and management quality from your words — directional), and model-generated commentary (fluent prose that may over- or under-state you — treat with caution). SCA management and interpersonal scoring especially require human calibration, so have a trainer, peer or study group review some of your recorded consultations against the simulator's verdicts rather than trusting the number. The examiner walkthrough is useful precisely because it exposes the author's reasoning for comparison.
Repetition risk and preserving unseen cases
A finite case library creates false fluency: replay the same 200 cases and you rehearse those consultations rather than building transferable consultation skill. Rotate across the full library, and preserve a reserve of never-attempted cases for the final fortnight so you can test whether your structure and timing hold on a cold scenario.
What it cannot test — and where iatroX honestly sits
Be clear about the gaps: non-verbal rapport, the full holistic judgement of a human examiner, examiner-to-examiner variability, and your own remote-consultation setup on the day. And be equally clear about iatroX's place: iatroX is a knowledge and clinical-reasoning platform, not a consultation simulator, so it does not replace Geeky Medics for the SCA. Its role in an SCA-preparation stack is the clinical knowledge the consultation rests on — the differentials to gather toward, the current UK management to offer, the prescribing and safety-netting that make a plan safe. A trainee who can consult fluently but reaches for outdated management will lose marks in the Clinical Management domain that no amount of communication polish recovers. Use the simulator to rehearse the consultation; use iatroX to keep the medicine inside it current.
A seven-day pattern for ST3 trainees
Monday: two Geeky Medics cases, strictly timed to 12 minutes, feedback reviewed against observable behaviours. Tuesday: one case plus a trainer or peer review of your recorded consultation against the simulator's scoring. Wednesday: a focused knowledge session in iatroX on the management and prescribing points your Tuesday review flagged as shaky. Thursday: two cases across under-practised Clinical Experience Groups. Friday: one case plus the examiner walkthrough studied for the author's reasoning. Saturday: a timed mini-circuit of three cases back to back, reviewed by domain. Sunday: rest, preserving unseen cases for later calibration. Geeky Medics rehearses the consultation; iatroX keeps the clinical content current; neither pretends to be the other.
A worked example: separating the three layers of SCA feedback
The SCA is scored across three domains, and an AI simulator's feedback blends what it can observe with what it can only guess — so a worked case is instructive. Suppose after a consultation the simulator returns domain scores plus: "Strong data gathering; management plan lacked complexity; good interpersonal skills." Weight each by observability. Data gathering is largely observable — did you cover the domains, screen for red flags, explore the patient's ideas, concerns and expectations? — so a strong score there is reasonably trustworthy and a weak one is actionable. "Management plan lacked complexity" is a genuine SCA domain (Clinical Management and Medical Complexity) but the model's judgement of it is inferred from your words, so treat it as a prompt to check whether your plan was actually current and safe, not as a verdict. "Good interpersonal skills" is the least reliable, because the Relating to Others domain turns substantially on non-verbal rapport the model cannot see.
The correct response is to fix what is observable, verify what is inferred, and largely discount what is unobservable — and to route the management comment specifically to a knowledge check, because a plan that "lacked complexity" often means the underlying management was thin or out of date rather than poorly communicated. This is precisely where iatroX's role sits: the simulator flags that the plan was weak; iatroX tells you whether the current UK management you should have offered is what you thought it was.
Why current management is the hidden SCA marker
It is tempting to treat the SCA as purely a communication exam and drill consultation structure endlessly, but the Clinical Management and Medical Complexity domain is where knowledge and consultation meet, and it is unforgiving of outdated practice. A trainee who consults beautifully and offers superseded management loses marks a communication polish cannot recover, because the examiner is judging the safety and currency of the plan, not just how warmly it was delivered. That is the argument for keeping a current-knowledge source in an SCA stack: the simulator rehearses the consultation, human observers judge the rapport, and a citation-grounded UK source keeps the medicine inside the consultation current — three jobs, and the simulator honestly owns only one of them.
Continue, supplement, switch or stop
Continue while your consultation structure, timing and management currency all improve. Supplement with human-observed practice for the interpersonal and management judgement the AI cannot fully assess. Switch only if the case mix or feedback proves unreliable for your needs. Stop replaying familiar cases in the final fortnight; calibrate on unseen scenarios under strict time and confirm the underlying management is current.
Frequently asked questions
Is Geeky Medics SCA enough for the MRCGP SCA on its own? No — it rehearses timed consultations and structure well, but the SCA is examiner-judged across management and interpersonal domains that AI automarking only approximates, so pair it with human-observed practice and current-knowledge revision.
Which MRCGP SCA component does Geeky Medics not reproduce well? The holistic examiner judgement of clinical management quality and interpersonal rapport, non-verbal communication, and examiner variability — its scores are directional, not a predicted result.
How many unseen Geeky Medics cases should I preserve for final SCA calibration? Keep a reserve of several never-attempted cases across different Clinical Experience Groups for the final fortnight, so your last practice measures transfer to cold, timed consultations rather than recall of rehearsed ones.
When should I stop using Geeky Medics and move to full practice circuits? When your structure and 12-minute pacing are reliable, shift the final stretch to timed multi-case circuits and human-observed consultations, using the simulator for warm-up and the preserved unseen cases for calibration.
How should I combine Geeky Medics with iatroX without duplicating practice? Use Geeky Medics to rehearse the consultation and iatroX to keep the clinical knowledge inside it current — differentials, UK management, prescribing and safety-netting — so a fluent consultation is also a clinically correct one.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; Geeky Medics SCA figures (200 cases, AI virtual patients, RCGP-mapped mark schemes, ~£10.99/month) are vendor-published — verify counts and price on the product page. MRCGP SCA format is per the RCGP. Disclosure: iatroX supports SCA knowledge preparation but is not a consultation simulator, and this audit says so plainly. Corrections via the feedback route on iatrox.com. References: RCGP Simulated Consultation Assessment pages (rcgp.org.uk); Geeky Medics SCA pages (geekymedics.com); related reading: why your Q-bank percentage is not your exam score.