Specification mapping
Each question lane starts with the subject, qualification, topic, board, and command-word pattern. The goal is not to make a plausible worksheet; the goal is to test the exact behaviour a mark scheme rewards.
Question quality
AI can help organise, explain, and route learning. It cannot be the final authority on exam practice. StudyVector treats question quality as a controlled content pipeline: syllabus mapping first, reviewed question logic second, adaptive delivery last.
Each question lane starts with the subject, qualification, topic, board, and command-word pattern. The goal is not to make a plausible worksheet; the goal is to test the exact behaviour a mark scheme rewards.
StudyVector separates live reviewed questions, draft candidates, and generated support material. A question is not treated as a live exam-style item unless its topic fit, wording, answer route, and explanation have been checked.
The answer feedback is tuned around likely mark loss: missing method, weak evidence, unclear units, poor command-word response, or an answer that is correct but not sufficiently precise.
The control layer
AI can help produce explanations, identify likely weak areas, rewrite feedback into clearer language, and route students to a better next task. It does not get to invent exam-board endorsement, assign official source labels, or push unchecked question stems into live practice.
Reviewed patterns
A good question is not just a topic label and an answer box. It needs the right command word, the right difficulty, the right route to marks, and feedback that helps a student recover the marks they missed.
A reviewed algebra item checks the board, topic strand, method route, final answer, units where relevant, and the most common mark-loss pattern: a correct idea with missing working.
Source and case-study questions are checked against command words, evidence use, explanation depth, and whether the answer actually earns analysis marks rather than listing facts.
Longer answers are framed around assessment objectives, module outcomes, reading evidence, and precision. Where StudyVector has uploaded or public source context, feedback shows that grounding level.
The product direction is simple: every explanation, quiz, feedback panel, and Battle Mode progression should make students answer better questions more consistently.