Most platform comparisons end with "it depends". This one ends with a workflow, because GPAtlas and iatroX slot together unusually cleanly: one is built for getting knowledge in, structured summaries, audio, consultation practice, and the other for proving it stayed, unseen assessment, misconception-level correction, delayed retesting and evidence. Here is the eight-step loop, followed by three weekly routines for the people most likely to run it.
The eight-step loop
Step one: read an Ocean clinical topic, or listen to it through Echo on the commute. Input first, and in whichever modality your week allows.
Step two: within a day or two, complete a short, unseen iatroX question set on the same territory. Unseen matters: the point is retrieval under mild pressure, not recognition of the page you just read.
Step three: review your errors in two piles. High-confidence errors, wrong and sure of it, are the valuable ones: each marks a live misconception that reading alone would never have surfaced. Ordinary errors mark thinner knowledge.
Step four: take the high-confidence pile to the Socratic Tutor. Its dialogue asks what you chose and why, names the actual misconception, and corrects it against guidance cited directly into NICE, CKS, SIGN or SmPC sources, which is a different event from re-reading the explanation.
Step five: retest after an interval. Spaced repetition schedules the topic's return; your performance on the return visit is the fact most CPD evidence never contains.
Step six: save the session as a CPD learning record. My CPD captures topic, date, active time, score, the specific errors, the corrections and the retest, automatically, as it happens.
Step seven: add a concise reflection in your own words. Two to four sentences: what you misunderstood, why it matters in your surgeries, what changes. The platforms scaffold; the sentences are yours, as current college guidance requires.
Step eight: at appraisal, export selected records rather than everything. A handful of full-cycle entries beats a hundred pages of microlearning, and the selection itself is good appraisal preparation.
Routine one: the AKT-and-SCA registrar
Weekly shape: three GPAtlas inputs, two topics read or listened plus one SCA case block with the marking sheet, and three short iatroX sets, two tracking the week's topics unseen and one purely adaptive, letting the engine roam the AKT blueprint for gaps your rotation has not touched. Tutor time goes to high-confidence errors only; everything else rides the spaced queue. One consultation practice session with humans stays sacred, because the SCA's interpersonal domain is not a self-study subject. Evidence output: the training portfolio gains one properly reflected full-cycle record a week, which quietly doubles as AKT preparation documentation.
Routine two: the post-CCT GP
One clinical theme per week, chosen from your actual case mix: read or listen in GPAtlas early in the week, one unseen iatroX set midweek, Tutor on the misses, retest lands the following week as the next theme begins. Quarterly, replace the theme with a breadth check: a broad adaptive sweep across the GP scope to find what independence has quietly eroded, feeding the next quarter's themes. Time cost: under an hour a week. Evidence output: roughly forty themed records a year plus four breadth reports, from which appraisal takes the best six.
Routine three: the locum GP
Capture-driven rather than theme-driven: the week's genuine point-of-care uncertainties, asked through Ask Expert or askiatroX as they arise, become the seed list. At week's end, run a targeted set across those seeds plus the adaptive engine's additions; Tutor and retest as above. The evidential point matters for locums specifically: this workflow demonstrates maintenance across the full GP scope from within a working life that no single practice's education programme covers, which answers the question locum appraisals always circle.
Why this beats either platform alone
Input without verification produces the familiar illusion of currency: pages read, podcasts finished, knowledge assumed. Verification without input eventually tests you into topics you have no scaffolding for. The loop pairs them: GPAtlas supplies structured, accessible input; iatroX supplies the unseen test, the diagnosed misconception, the delayed proof; and the record of the whole cycle demonstrates more than a certificate of attendance alone, because it shows what you did not know, how it was corrected, and that the correction held.
Frequently asked questions
Do I need paid tiers of both platforms?
To run the full loop weekly, realistically yes eventually, though both free tiers cover a genuine trial: GPAtlas's free tier with its limited AI credits for input, iatroX's free tier with askiatroX and selected full banks for the assessment half (both last checked August 2026). Trial the loop free for a fortnight before spending anything.
Does the same loop work with other content sources?
Entirely: swap GPAtlas for a guideline, a course or a journal club and the verify-correct-retest half is unchanged. This article pairs the two because they fit well, not because the loop is proprietary.
Which platform's CPD capture should hold the record?
Either works; avoid duplicating. Our bias is for the assessed record as the exportable artefact, with reading and listening noted inside its reflection, one entry per cycle, not three.