Journal

A commit-first workflow that spends UWorld's capped UAsk prompts where they earn their keep on Step 3, keeps its CCS simulator a separate scheduled task, and protects the independe...

Medibuddy's adaptive engine and Topic Mastery scores optimise practice performance, not MRCP Part 1 readiness — here is what each metric measures, where the adaptive feed biases yo...

Pastest's AI Tutor meets an exam where UK guideline currency and critical-appraisal reasoning decide the result — this audit gives GP trainees a reproducible rubric for testing its...

PassMedicine can drill MRCGP AKT knowledge, but its dashboard can't see the evidence-interpretation and organisational thirds that decide the exam — this audit explains what each m...

A commit-first workflow that spends UWorld's capped UAsk prompts where they earn their keep on Step 2 CK — reasoning prompts, a synthesis-drift check against the written explanatio...

A commit-first workflow for Pastest's AI Tutor on the MSRA: reusable prompts that build reasoning rather than deliver answers, a hard boundary around the Professional Dilemmas pape...

Twenty-five reusable prompts that turn Pastest's AI Tutor from an answer key into a reasoning partner for MRCP Part 1 — grouped by error type, with an anti-answer-leak rule and a w...

UWorld pairs the market's benchmark Step 3 explanations with its own CCS simulator and the UAsk assistant — this review tests the AI layer's grounding, its fit to a management-heav...

UWorld added UAsk, an AI assistant with prompt limits, to the most trusted Step 2 CK explanations in the market — this audit tests what the AI layer grounds, where it can drift fro...

Vera Health searches and synthesises the literature in real time. UpToDate relies on a long-established, expert-authored editorial process. These are genuinely different models, wi...

Vera reports 97.5% on USMLE, 84.9% on NEJM-AI and 62.2% on MedXpertQA. These are genuinely strong, company-reported numbers, and it is worth being precise about exactly what they d...

Vera says it searches more than 60 million papers, guidelines and care pathways. A genuinely large corpus is valuable, but corpus size and answer quality are not the same thing, an...