EdAI Prep · For candidates

Prepare against the same validated corpus your exam is built from.

Spaced-repetition study and calibrated practice, scoped to what your board or jurisdiction actually requires. Every question traces to a physician-certified source you can open mid-question.

The question on the right is a real production item. Answer it.

Start preparing See the exams

Practice · Question 1 of 6

Delayed / precocious puberty · recall

A 9-year-old girl is referred to a general paediatric outpatient clinic with concerns about short stature. She is on the 2nd centile for height, with no dysmorphic features and normal systemic examination. The clinician wishes to assess skeletal maturation using a plain radiograph of the left hand and wrist. Which of the following best describes the Greulich-Pyle method of skeletal age assessment?

The exams

Five licensing exams live today. Each in its own jurisdiction’s lane.

The architecture works for any jurisdiction with a blueprint: blueprint in, validated preparation out. Content is retrieved from a jurisdiction-scoped slice of the corpus, so the right drug names, reference ranges, and referral thresholds stay in the right exam.

USMLE

$299 / year

Steps 1, 2 CK, and 3 · United States

One subscription covers all three steps. NBME register per step, through independent-practice management stems for Step 3.

Prep for USMLE

MCCQE Part I

CA$419 / year

Medical Council of Canada · Canada

English and French question content, weighted to the MCC blueprint.

Prep for MCCQE Part I

UKMLA AKT

£229 / year

UK Medical Licensing Assessment · United Kingdom

UK drug names, reference ranges, and referral thresholds, in their own lane.

Prep for UKMLA AKT

AMC CAT MCQ

AU$429 / year

Australian Medical Council · Australia

Community-practice reasoning, weighted to the AMC’s published patient-group blueprint.

Prep for AMC CAT MCQ

PRES Level 2

€269 / year

Irish Medical Council · Ireland

Including professionalism and ethics framed to the four domains of Good Medical Practice.

Prep for PRES Level 2

ENARM

In development

Examen Nacional de Aspirantes a Residencias Médicas · Mexico

In active development. The Spanish-language clinical-case engine and the exam’s official CIFRHS content weighting are built; it goes live once the remaining source content is in place.

About ENARM on Prep

Clinical-skills examinations (NAC OSCE, PLAB 2, NZREX Clinical) are coming and not yet live. When something is not ready here, it says so.

How you study

Four ways in. One corpus underneath.

Practice blocks

Twenty questions, then your due flashcards.

Free-form practice on demand. Every question grades instantly with a full explanation and a provenance badge naming who wrote and who certified the source behind it. The block finishes with a batch of due flashcards seeded around the concepts you just practiced.

Single-best-answer items in your exam’s register
The certified source review one click away, streamed as the actual PDF
Targeted drilling: arrive from a weak concept and the block aims at it

Clinical cases

One coherent patient, several linked decisions.

For exams built on case testlets, a shared vignette carries multiple linked questions. Cases are checked for clinical coherence before you ever see them: a patient course a real physician would find implausible is rejected, not served.

Shared vignette with linked single-best-answer items
Per-item explanations after you submit the whole case
Exam-native axes, like ENARM’s level-of-management category, revealed only after grading

Mock exams

Full length, timed, under real conditions.

A sticky timer, a question navigator with mark-for-review, and auto-submit at zero. The report breaks your score down by content area and by cognitive level, names your weakest and strongest areas, and updates your study plan from the attempt.

One included attempt per subscription, more by credit
By-area and by-cognitive-level score breakdown
Item-by-item answer review after the report

The study plan

Your test date, your hours, your weak concepts.

A wizard takes your test date, minutes per day, and days per week, optionally calibrates with a short adaptive quiz, and builds a day-by-day program. Guided sessions walk the exact mix the plan wants today, and the plan re-projects itself after every session.

Four methodologies: spaced-repetition first, qbank mastery, foundational curriculum, or high-yield cram
A coverage contract showing exactly how every topic was weighted, and why
Pace, streaks, focus-area drills, and calendar export

Flashcards

FSRS scheduling that projects retention to your actual test date, with export to a real Anki .apkg file.

Study guides

Synthesized per topic from the certified corpus, with listen-aloud controls, offline download, and a sources list naming every review used. No source content, no guide: it says so rather than fabricating.

The review library

Browse and read every certified review behind the question bank, independent of any question.

Progress

Per-exam mastery by system, weakest first, with one-click drilling into weak areas.

Verified before it reaches you

Every item passes three gates. Or it never ships.

Generation from a certified source is the start, not the point. An AI system turning a certified claim into a five-option question can still introduce errors in the packaging. So every generated item runs a stack of automated integrity checks before a learner ever sees it.

Every gate fails safe: when the system is uncertain, it regenerates, trims, or withholds. And the live pool is re-audited continuously, not just checked once at generation time.

Vignette coherence
A separate review pass, never the model that wrote the question, reads each multi-question case and asks whether a real physician would find the patient course clinically plausible. Incoherent cases are rejected or trimmed.
Grounding integrity
For every explanation of why an option is wrong, an independent judge verifies the cited certified claim actually establishes it. Explanations that merely sound authoritative do not pass.
Structural defensibility
If the certified evidence cannot cleanly support one single best answer, the item is pulled rather than shipped with a manufactured key.

What we will not tell you

The honest version of every number.

Preparation progress, never a pass probability

The readiness metric is blueprint coverage and measured retention, shown with a confidence band: an early "keep going" range first, an approximate figure when the evidence supports one, a precise one only when it is earned. Nobody can honestly promise you a pass probability, so we do not.

Guides that refuse to fabricate

Ask for a study guide on a topic with no certified source content behind it and the product tells you so. It never synthesizes a guide from nothing.

Drills that admit when they cannot drill

If a weak concept cannot be targeted, the app quietly gives you an ordinary block rather than falsely claiming to be drilling it.

Translations ship after human review, not before

French is live for MCCQE. A machine-drafted Spanish layer exists across the platform today and stays gated until it clears physician review. Verified, then released.

See it

Watch a study session

The adaptive plan, a clinical case with sourced explanations, and the spaced-repetition review flow.

Demo film in production