The MRCGP examines the same specialty through two very different lenses. The AKT is a written test of applied knowledge: clinical management per current UK guidance, evidence interpretation and statistics, and the administrative machinery of NHS general practice. The SCA assesses simulated consultations: data gathering, management in partnership, and the relating skills that make ten minutes work. Preparing for both with one tool is a category error; the strongest 2026 approach assigns each exam its machine.
The AKT: guideline-grounded practice at volume
The AKT is written against the guidance GPs actually use, which makes source-grounded question practice the core engine. iatroX fits this exam unusually well: the AKT bank is curated and adaptive, spaced repetition schedules your weak domains, timed mocks train the pacing that catches trainees out, and askiatroX resolves the "what does NICE actually say" questions that AKT revision generates daily, with direct citations into NICE, CKS, SIGN and SmPC sources. The Socratic Tutor engages on missed questions, which for the AKT often means surfacing that you learned an old threshold or a hospital habit rather than current primary-care guidance. Statistics and evidence-interpretation items respond to the same treatment: practise, miss, get interrogated, retain.
For a trainee whose revision happens between surgeries, the native apps matter: twenty adaptive questions over lunch, reviews on the commute, the planner absorbing the week an out-of-hours block destroys.
The SCA: rehearsing the consultation
The SCA needs a different rehearsal room. Geeky Medics' virtual patients and AI examiner feedback build the component skills, structured gathering, shared decision-making, safety-netting, and dedicated SCA simulators and case courses go deeper into the exam's specific format and marking domains. General-purpose voice AI can add cheap extra reps: have it play a patient with an undifferentiated presentation and hold you to twelve minutes. None of these grades you like the real thing; all of them beat rehearsing in your head.
The underrated SCA preparation, though, is daily practice itself, consciously debriefed. Every surgery is a bank of cases; the trainees who improve fastest treat them as such, reflecting on the consultation that went sideways and feeding the learning need into their revision loop.
The bridge between daily work and both exams
This is where an integrated platform earns its keep for GP training specifically. The same clinical questions you look up between patients are AKT content; the same cases that stretch you in surgery are SCA material; and both are CPD. iatroX's loop is built for exactly this circulation: askiatroX answers the in-surgery question with cited guidance, the AKT bank turns the topic into retrievable knowledge, and My CPD captures the reflection so the learning event also serves your portfolio. One subscription at £29 monthly or £99 annually spans the AKT bank, Tutor, planner and mocks, and remains useful after the exams as the daily reference and CPD layer.
The combined season
A workable shape: AKT-first months of daily adaptive sets and weekly timed blocks, SCA skills ticking over with one rehearsed consultation a week; then inverted proportions as the SCA approaches, with the bank in maintenance mode. Throughout, let real surgeries feed both machines. The MRCGP is, in the end, an exam about the job you are already doing; the best AI setup is the one that closes the loop between the two.
Frequently asked questions
AKT first or SCA first?
For most trainees, AKT first, and not only because eligibility timing usually points that way: the knowledge the AKT forces into place is the raw material the SCA's management conversations draw on, and consulting fluently about guidance you have not yet internalised is precisely the failure mode examiners describe. Keep SCA skills ticking over lightly during AKT season, then invert the proportions, rather than treating them as strictly sequential.
How do I use real surgeries for exam preparation ethically?
Reflect, never record: no patient details enter any AI tool, ever, and the raw material is your anonymised reasoning, not the patient's story. The compliant loop is straightforward and powerful: after a stretching consultation, write what you gathered, decided and safety-netted; let a Socratic tutor interrogate the reasoning; convert the learning need into question-bank practice; log the reflection as CPD. The patient contributes nothing identifiable, and your appraisal gains an entry.
What actually works for the statistics and evidence questions?
Practice over reading, as everywhere else, but with one addition: make the tutor explain your wrong answer in words, not formulae. Most AKT statistics errors are concept confusions, absolute against relative risk, what a confidence interval is actually claiming, and a tutor that asks what you thought the number meant fixes the confusion in a way a worked calculation does not. Ten such interrogated questions beat an evening with a statistics chapter.
Is the platform still useful after CCT?
That is rather the design test, and the honest answer for iatroX is yes by construction: askiatroX and the calculators serve daily surgery, My CPD serves appraisal, and the exam machinery goes quiet until you choose another mountain.