A missed question is worth more than a correct one, but only if you process it properly. This is a repeatable routine for doing exactly that with Pastest's AI Tutor on the UKMLA AKT — the implementation companion to our Pastest UKMLA audit. The goal is to make every miss a grounded, verified, dated learning event instead of a fluent explanation you nod along to and forget.
The routine in one line
Commit, interrogate, verify, record. Four steps for every genuine miss, and the order is the whole method: commit before the tutor opens; interrogate your reasoning rather than requesting the answer; verify anything behaviour-changing against a named source; record the corrected rule with a review date.
Step one: commit
Write your answer and a one-line rationale before Tutor mode opens. This preserves retrieval practice and gives the AI something real to correct. On the AKT — single-best-answer under time pressure — the skill is fast discrimination, and you cannot build it by reading explanations to questions you never actually attempted.
Step two: interrogate
Use prompts that force reasoning, not delivery. "Name the discriminating feature between my choice and the key — don't just tell me the answer." "Why is my distractor wrong specifically?" "What single change to the stem would flip the answer?" "Give me the underlying rule as something I could apply to a different presentation." These turn a wrong click into a transferable rule, which is the only kind of learning that survives to exam day.
Step three: verify — with a jurisdiction check
For anything that will change how you answer future questions, force a source: "What guideline supports this, and what is its date? Does it reflect current UK practice?" This matters doubly for internationally-trained candidates, whose existing knowledge is the interference pattern — the answer that feels right may be a home-country default. Where you want a fast, cited UK answer, Ask iatroX grounds its responses in NICE, CKS, SIGN, SmPC/eMC and NHS content, so the check takes a minute. Treat any confident, sourceless management claim as unverified until you have checked it.
Step four: record
Four lines per miss: the incorrect rule, the corrected rule, one transfer question, one review date. The record is the durable asset — a personal, dated list of exactly the misconceptions you have corrected, ready for spaced review. A month of disciplined records is worth more than the entire tutor transcript that produced them.
The weekly verification sample
Independently check five random tutor outputs against official or primary sources each week and log discrepancies. This keeps your trust calibrated as Pastest iterates its AI, and it is the same discipline the audit formalises. A clean log means you can lean on the tutor; a dirty one means you caught the drift before the exam did.
A seven-day pattern — with an IMG lens
Monday: 40 Pastest questions in weak map areas, commit-first, interrogation prompts on misses. Tuesday: 30 questions plus UK-convention review for the jurisdiction gaps your verification step keeps flagging. Wednesday: a timed, unseen 50-question mixed block in iatroX's free UKMLA bank, no tutor. Thursday: misconception-record review; weekly verification sample. Friday: 40 questions on professional and low-volume content, timed. Saturday: full timed simulation, same-day review. Sunday: rest. Pastest explains and interrogates; iatroX measures and verifies; your records are the output.
A worked missed-question walkthrough
The four-step routine is easiest to trust when you see it run once. Take a question where you chose "urgent referral" and the key was "safety-net and review in 48 hours" for a presentation the guideline manages in primary care first.
Commit: before opening the tutor you have already written "urgent referral — thought the red-flag features met the two-week-wait threshold". That committed rationale is what makes the next step diagnostic rather than decorative.
Interrogate: "Without giving the answer, which specific feature in this vignette sits below the referral threshold, and what is the safety-netting alternative?" The tutor identifies that the red flag you keyed on is only a referral trigger when combined with a second feature that is absent here — a discrimination you will now recognise across a whole class of questions.
Verify, with the jurisdiction check: "What UK guideline sets this threshold, and what is its date?" You open NICE or CKS — or ask a citation-first system for the source directly — and confirm the threshold and its currency, because referral criteria are exactly the content the AKT keeps current and the content an older-trained model gets confidently wrong.
Record: four lines — incorrect rule ("this red flag alone warrants two-week-wait referral"), corrected rule ("this red flag warrants referral only with a second feature; otherwise safety-net and review"), one transfer question, one review date. That record, not the tutor's paragraph, is what you revise from next week.
The whole cycle takes five minutes and produces a dated, testable correction. A candidate who runs it on every genuine miss finishes the revision period with a personal catalogue of exactly the thresholds they used to get wrong — which is worth more than any number of freshly-answered questions.
Why the jurisdiction check matters most here
Of the four steps, verification with a jurisdiction lens is the one to protect fiercely on the UKMLA, and doubly so for international graduates. The AKT's management answers track current UK practice, and the two ways to lose those marks are a stale answer (the guideline moved) or a transplanted one (you applied another system's default). Both are invisible in the moment — the wrong answer feels as fluent as the right one — and both are caught only by forcing a named, dated, UK source. Make that non-negotiable for any answer that would change how you respond, and the single largest category of avoidable applied-knowledge-test errors closes.
One thing this routine does not cover
Be clear about the boundary: this workflow, and Pastest's bank, address the applied knowledge test only. The UKMLA also includes a separate clinical and professional skills assessment — the CPSA — which is a practical, station-based examination that no question bank and no AI tutor rehearses. Nothing in this routine prepares you for it, and it would be a mistake to let a strong AKT question-bank position lull you into neglecting CPSA preparation, which needs supervised practical practice, examination technique and communication rehearsal on entirely different terms. Treat the two components as separate projects with separate methods; this routine is the whole of one and none of the other.
Continue, supplement, switch or stop
Continue while the routine holds and verification stays clean. Supplement with unseen timed blocks when explanation quality is high but measured performance is flat. Switch only on logged failures. Stop opening the tutor during blocks in the final fortnight.
Three mistakes this routine is designed to stop
First, skipping the commit step because the tutor is right there. Every consultation before you have written an answer and rationale converts a retrieval attempt into an assisted read — it feels efficient and it quietly deletes the practice effect. The written rationale is not optional; it is what makes the whole routine work. Second, accepting fluent, undated answers on management questions. The AKT rewards current UK practice, so any behaviour-changing claim without a named, dated source is a guess in the costume of an answer — force the jurisdiction check every time. Third, letting the tutor transcript substitute for the misconception record. The transcript is the input; the four-line dated record is the durable asset you actually revise from. A candidate who keeps the records finishes with a personal catalogue of exactly the thresholds and conventions they used to get wrong — which is worth more than any number of freshly-answered questions, and is the thing an always-available explainer most tempts you to skip.
Frequently asked questions
Is Pastest enough for UKMLA on its own? Its bank and AI can anchor AKT preparation, but only alongside a content-map audit, unseen timed blocks and CPSA preparation the bank does not cover.
Which UKMLA component does Pastest not reproduce well? The map's professional and ethical breadth and the unseen timed conditions of the paper — plus the retrieval practice an always-available tutor threatens unless you commit first.
How should I verify Pastest AI answers for UKMLA? Force a named, dated, UK-specific source on every behaviour-changing claim, check it directly or via Ask iatroX, and audit five random outputs weekly with a written log.
When should I stop using Pastest and move to mixed mocks? When map coverage is complete, accuracy is stable and pace is on budget — the final two to three weeks, tutor closed during blocks.
How should I combine Pastest with iatroX without duplicating practice? Pastest for commit-first drilling and reasoning prompts; iatroX (free for UKMLA) for unseen measurement, Socratic repair and cited verification.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; Pastest features vendor-published and evolving; MLA content-map dates per the GMC. Disclosure: iatroX offers a free competing UKMLA bank and Socratic Tutor. Corrections via the feedback route on iatrox.com. References: GMC MLA content map (gmc-uk.org); Pastest product pages (pastest.com); related reading: the Pastest UKMLA audit and why your Q-bank percentage is not your exam score.