Marking is the single biggest drain on teacher time in most departments. It is also the work that most directly shapes what happens next in the classroom. When a set of mock papers takes a fortnight to turn around, the feedback lands too late to change anything, and the staff who marked it are too stretched to plan the response. This is the problem ilmino's AI marking is built to solve: return accurate, consistent marks in a fraction of the time, while keeping teachers fully in control of every score.
The principle underneath everything below is simple. AI assists, teachers stay in control. Teachers can see every mark, override any score, and have the final say.
The marking workload problem for departments
Across a typical department, marking competes with planning, intervention and the rest of the working week. The cost is not only the hours. It is the knock-on effects:
- Feedback arrives late, so students have moved on before they act on it.
- Marking standards drift between classes and between markers, which makes results harder to compare.
- Heads of department cannot see where a cohort is struggling until the marking is finished and collated, which is often too late to intervene.
ilmino attacks the time cost first, then uses the time it frees to improve consistency and insight. Each submission is marked in seconds, and the teacher's role shifts from doing every mark by hand to checking and confirming the ones that matter.
How the split review view works
When a submission is marked, ilmino opens it in a split review view. The student's paper sits on one side and the AI marking sits on the other, so a teacher reviews both together without flicking between documents.
For every question the teacher sees three things:
- The mark awarded.
- A mark-scheme-aligned reason for that mark, for example "the correct term is chemical".
- A confidence score for the mark.
This turns review into a fast, structured read rather than a fresh marking exercise. The teacher is checking a decision that has already been made and justified, not starting from a blank script.
Mark-scheme-aligned reasons per question
The reason attached to each question is the part that makes the marking auditable. Rather than a bare number, every mark carries a short justification tied to the mark scheme, so a teacher can see at a glance why a point was awarded or withheld. That matters for three audiences at once: the teacher checking the paper, the student reading the feedback, and the head of department moderating across classes.
Typed and handwritten work, marked step by step
Departments do not mark in one format. Written papers, worked calculations, extended answers and quick recall tasks all sit side by side. ilmino marks across typed and handwritten answers, so a maths working, a science definition and an essay response can all go through the same process.
For handwritten working, ilmino reads the working and marks method step by step, not just the final answer. A student who takes a valid method to a slip at the last line is credited for the method, exactly as a teacher marking by hand would credit it. This is what makes automated marking usable for subjects where the method is the point, not an afterthought.
Confidence scores and where-to-improve summaries
The fastest way to waste the time you have saved is to re-check every mark equally. ilmino is built so teachers can spend their attention where it counts.
- Confidence scores flag the marks the AI is least sure about, so teachers know which ones to check first. High-confidence marks can be scanned quickly; lower-confidence marks get proper scrutiny.
- Each submission carries a "where to improve" summary, so a teacher can see the shape of a student's performance without reading every line, and can point the student at the right next step.
Together these turn a full re-mark into a targeted review. The teacher's expert judgement goes to the borderline decisions rather than the obvious ones.
The accept or override workflow
This is where teacher control is concrete rather than a slogan. For every question, the teacher can:
- Accept the AI score, or
- Override it with a manual score and comments.
As the teacher adjusts marks on the paper, the manual score updates to reflect their decisions. Nothing is locked. The AI's proposal is a starting point that the teacher confirms, corrects or replaces, and the final mark on the paper is the teacher's mark.
Why keep judgement final? Because a mark scheme never covers every valid response, context matters, and the teacher knows the student and the cohort in ways an automated system does not. ilmino is designed to do the heavy, repetitive part quickly and to surface its reasoning, so the teacher can apply judgement efficiently rather than from scratch. Results and feedback publish to students only once the teacher is satisfied.
Consistency across a whole department
The benefit of a shared marking process grows when a whole department uses it. Because every submission is marked against the same mark scheme with the same reasons attached, the baseline is consistent before any human moderation begins.
That supports two things departments already care about:
- Standardisation. Different classes are marked to the same standard, so results are comparable across sets and teachers.
- Moderation. When markers meet to moderate, they are reconciling small differences against a consistent baseline and a visible reason per mark, not arguing from scratch over raw scripts.
Feeding analytics and intervention signals
Once results and feedback publish, they feed department and trust analytics. Marks and the "where to improve" summaries roll up into a view of where a cohort is strong and where it is struggling, question by question and topic by topic. For a head of department or a trust lead, that turns marking from a backward-looking chore into an intervention signal: the data shows which topics to reteach and which students to target, and it shows it early enough to act.
A note on responsible use
Automated marking has to be handled responsibly, and the design reflects that:
- Teacher oversight is built in, not optional. Every mark is visible, every reason is shown, and every score can be overridden. The teacher confirms what publishes to students.
- Transparency over black boxes. Mark-scheme-aligned reasons and confidence scores mean a mark can always be explained and questioned, which matters for fairness and for appeals.
- Data handling. Student work and results are handled as the sensitive records they are, and used to support marking, feedback and analytics within the institution rather than to replace professional judgement.
The aim is not to remove teachers from marking. It is to remove the repetitive burden while keeping teachers firmly in charge of the decisions that affect students.
How to roll it out in a department
A department does not need to change everything at once. A sensible sequence:
- Start with one assessment. Choose a single mock or end-of-unit test and run one class through the split review view to see how accept and override feels in practice.
- Agree a review routine. Decide as a team to check low-confidence marks first, then spot-check the rest. Standardise how overrides and comments are used.
- Use it for the next moderation. Bring the consistent baseline and per-question reasons to your moderation meeting and compare against your usual process.
- Widen to the department. Once the routine is comfortable, extend it across classes so standardisation and analytics cover the whole cohort.
- Act on the analytics. Use the department and trust views to target reteaching and intervention while the feedback is still fresh.
Marking will always need teacher judgement. ilmino's job is to give your department that judgement back its time, and to make every mark faster to check, easier to justify and more consistent across the team.
Ready to see it on your own papers? Book a demo and we will walk your department through the split review view, the accept and override workflow, and the analytics behind them.
