How it works
Three steps, start to finish.
- 1
The scheme comes along
Marking schemes come from the paper, part by part, so 2(b)(i) is marked as 2(b)(i), with the question's context.
- 2
AI suggests, with reasons
Each answer gets a mark and a sentence on why: what it got and what it missed. Structured output, so marks never get lost in a reply.
- 3
Release when you're sure
Go answer by answer or question by question across the class. Nothing reaches students until you release it.
What you get with ai marking.
Typed and scanned
Works on online answers and on scanned paper scripts.
Part by part
Marks follow the paper's own numbering, with the question's context.
Office and code too
Word formatting, Excel formulas and slides are described for AI; code is run against tests (practicals).
Reasons, not just numbers
Every suggestion says why, so a change takes seconds.
People release
AI never changes a released mark by itself.
Marked lists stay yours
Teachers see only the batches they teach in marking lists.
How it compares.
Fair, sourced comparisons with the tools schools weigh this against, checked on their own sites.
Questions about ai marking
Can AI mark handwriting?
Yes. Scanned pages are read by Google Gemini's vision model, then marked against the scheme. A teacher checks every suggestion before release.
Does AI give the final mark?
No. AI suggests; a teacher agrees, changes or writes their own, and only released marks reach students.
Which questions can it mark?
Short and long answers, diagrams by description, MCQs automatically, Office files by their contents and code by running tests.
Is students' work used to train AI?
SparkMentis calls Google's Gemini API to mark; it doesn't train models on your students' work.
Goes well with.
Across SparkMentis
See it with your school.
Tell us about your school and we'll walk you through a term with your own classes and papers.