Built for students who cannot get started.

Papers are read, not invented

AceUrExam works from authentic exam board material — Cambridge (CAIE), Pearson Edexcel and AQA papers across exam years 2018 to 2026, covering sessions, paper numbers and variants. Nothing in a pack is a generated imitation of an exam question. When you open a topical pack, every question in it was sat by real students in a real session, and the mark scheme beside it is the official one published for that paper.

Questions are found, then checked

Finding where one question ends and the next begins is the hard part of turning a PDF back into questions. AceUrExam segments each paper by its layout, then verifies the result rather than trusting it. A topic match has to be grounded in the question's own text: matches the model cannot support with a quote from the paper are dropped rather than shipped. The effect is a pack that is smaller than a naive keyword sweep would produce, and far less likely to contain a question that is not really about your topic.

Mark schemes are paired by verified identity, not by filename. A mark scheme is only attached once the pipeline has confirmed it belongs to that exact paper, session and variant — so the answers you mark against are the answers for the questions you actually attempted.

Mock papers are shaped, not shuffled

A pile of real questions on the right topics is not a mock paper. Before assembling one, AceUrExam reads real papers of your syllabus and paper number to learn that paper's shape: which kind of topic tends to open it, how the mark tariff grows through it, where the long multi-part question sits, roughly how many questions a session runs to. Your questions are then fitted into those positions. The result sits like the paper you will actually be handed, which is the difference between practising questions and practising an exam.

Difficulty is scored, then corrected by students

Every question found on a topic can be scored from 1 to 10 by an examiner-style model — one-step recall at the bottom, unfamiliar framing and multi-topic synthesis at the top. That first score is a model's opinion, so it is not left alone: students who attempt a question can mark it as genuinely hard or manageable, and those votes progressively correct the ranking. Over time the order reflects what a cohort found difficult rather than what a model predicted.

Where AI is used, and where it is not

AI reads and organises; it does not write your practice material. It decides which real questions are about your topic, maps mark scheme answers onto them, ranks difficulty, and writes the solution guide that explains how each mark is earned. The questions and the mark schemes themselves are always the board's own. Revision notes are the one place the writing is AI's — and they are labelled as such in your library.