The problem is more than counting marks
A pile of answers represents several jobs at once. You need to read what each student means, compare it with the question, allow partial credit and decide which misunderstanding deserves feedback. The difficult part is often the repeated judgment, not the arithmetic.
An unexplained AI score does little to help. If you have to reconstruct why it awarded a mark, you have simply acquired another answer to investigate. Useful assistance needs to show a basis you can review.
How Quesly helps: a suggestion with marking points
For written answers, Jev drafts marks and reasons that teachers can inspect alongside student work. Quesly keeps those suggestions separate from the final marks you enter and save. MCQs take a different route: they are checked against one fixed answer key.
This gives you a starting point for review, rather than asking you to accept an automated verdict. It is designed to support the marking workflow; we have not established a measured reduction in marking time or independently validated grading accuracy.
The answer may be right in different words
Consider an illustrative two-mark question about how faster and larger vibrations change sound. One marking point might concern higher pitch, and the other louder sound. A student who explains only pitch has shown part of the required understanding.
Now consider a student who writes “a higher-pitched sound” instead of the wording in the draft answer key. The meaning matters more than a phrase match. Read the response before deciding whether the suggested mark recognises that understanding, and correct defective criteria rather than penalising the student for them.
Keep the final review deliberate
Finish by checking missing marks, entering zero where appropriate and saving your decisions before releasing results. Students see released results, giving you time to correct the review before feedback is shared.
Start with a small assignment and note where suggestions help or mislead. The goal is a review process you can stand behind—not a faster-looking number that hides an unchecked judgment.
