The tools question, what should I be using to learn?, has no stable answer, because the underlying problem keeps changing. The student's problem is acquisition; the FY1's is safe application; the registrar's is exams plus depth; the consultant's is maintenance against invisible decay. Map the stages honestly and the right stack at each becomes almost obvious, along with the one thread that should run through all of them.
| Stage | Primary learning problem |
|---|---|
| Medical student | Acquire a vast knowledge base |
| FY1 | Apply knowledge safely under pressure |
| FY2 | Develop breadth and independence |
| Specialty trainee | Pass exams while deepening expertise |
| New GP or consultant | Maintain competence independently |
| Established clinician | Identify hidden gaps and drift |
| Educator | Teach, and demonstrate development |
Medical student: acquisition
The problem is volume, and the stack should be comprehension-first with retrieval built in: a video or library layer for understanding (Osmosis, Zero to Finals, the school's own teaching), question banks early rather than late, because testing is learning, not just measurement, and spaced repetition from the start. The commonest student error is treating questions as a final-term activity; the evidence says they are the curriculum's engine from year one.
FY1: safe application
The problem changes shape entirely: knowledge exists, and the job is deploying it under pressure without harm. The stack becomes reference-and-reflex: fast cited guidance access (askiatroX, CKS, local pathways), calculators always, a small memorised safety core drilled to reflex, and the capture habit that turns the week's real questions into next week's retrieval practice; the full setup is at /blog/starting-fy1-2026-resources-on-my-phone, and the memorise-versus-look-up line at /blog/what-should-an-fy1-know-without-looking-it-up.
FY2 and the F3 years: breadth
Rotations widen, supervision lightens, and the problem is breadth with no one testing you. The stack adds deliberate self-assessment across the widening scope, adaptive questions finding the gaps rotations skipped, and, for the majority now taking years outside training, the self-built structure that replaces the programme's: /blog/what-to-do-after-f2-no-training-number.
Specialty trainee: exams plus depth
The classic double problem: blueprinted exams to pass while genuine expertise deepens. The stack centres on a curated adaptive bank for the exam (with mocks and a planner), a Socratic layer interrogating misses rather than explaining past them, and the specialty's own literature for the depth the exam only samples. This is the stage the exam-tutor market serves best, and choosing by exam is the whole game.
New GP or consultant: independent maintenance
Exams stop; decay starts, invisibly. The problem is maintaining a large knowledge estate with no external forcing function, and the stack becomes the maintenance loop: periodic broad assessment finding drift, targeted correction against current guidance, spaced retesting, and, since appraisal now wants evidence, records that document the whole cycle, the loop described at /blog/how-doctors-actually-learn-after-medical-school, with the CPD machinery at /blog/first-gp-appraisal-after-cct.
Established clinician: hidden gaps
A decade in, the problem sharpens: what you know, you know well; the danger is what changed without telling you and what you never knew you missed. The stack's centre of gravity shifts to adaptive breadth-sampling, deliberately testing the neglected corners of scope, guideline-change awareness, and the honest audit that a high-confidence error is the most valuable finding available at this stage, not an embarrassment.
Educator: teach and demonstrate
The final stage's problem is double again: keeping your own edge while building others', and evidencing both. Teaching itself is the stack's secret component, explanation being retrieval practice at its most demanding, joined by education credentials (PGCert onward), feedback systems that capture teaching evidence automatically, and the same maintenance loop underneath, because the educator's credibility is their own currency.
The thread through all seven
One mechanism appears at every stage in different clothing: retrieval, spaced, with correction. The student's flashcards, the trainee's bank, the consultant's maintenance loop and the educator's teaching are the same engine at different loads, which is why the platforms worth long-term loyalty are the ones built on that engine rather than on any single stage's content. Stage-match the stack; keep the engine.
Frequently asked questions
Should tools be replaced wholesale at each stage?
The content layer, often; the engine layer, ideally never, since continuity of performance data is what lets a system find your drift years later. Prefer platforms that span stages over per-stage purchases where quality allows.
Which transition do doctors most under-prepare for?
Trainee to independent practice: the exam scaffolding vanishes and nothing external replaces it. The doctors who thrive built their maintenance loop in the final training year, not the fifth independent one.
Where do college and society resources fit in this model?
As stage-specific content layers riding the same engine: curriculum and updates from the college, retrieval, spacing and correction from your system. The pairing is complementary by design, not competitive.