Grade
Your rubric applied consistently across the cohort, borderline cases surfaced first.
With AI assistance on, essays and speaking receive draft scores and rubric-anchored explanations. The marker reviews, overrides, signs off. Inter-rater agreement is visible. AI never makes the final grade decision.
Marking is a bottleneck. Removing it cannot mean removing the marker.
A programme that marks thousands of essays a term has a real operational problem. Feedback arrives too late to inform learning. Marker consistency drifts across the cohort. Higher-order feedback is the first casualty when time is short.
AI marking that replaces the marker creates a different problem: it puts a high-stakes academic decision behind a model that nobody on staff can fully explain, and removes the human accountability that the rest of the institution depends on. That is not a fix — it is a transfer of risk.
AI as a documented second rater, with the marker still signing off, gets you the operational improvement without the transfer of risk. That is the pattern NUADU implements.
Your rubric applied consistently across the cohort. The marker decides.
The marker decides.
Your rubric applied consistently across the cohort, borderline cases surfaced first.
AI-assisted marking — spoken responses
With AI assistance on, essays and speaking receive draft scores and rubric-anchored explanations. The marker reviews, overrides, signs off.
Objective scoring
For MCQ, matching, ordering, code and other objective formats, scoring is deterministic — partial credit, synonym acceptance and tolerance ranges as configured. Distractor analysis available for item review.
Marker review and sign-off
The marker sees the cohort with borderline cases at the top. Reviews, edits, overrides where needed. Signs off. Inter-rater agreement, override rate and time per response are visible to QA. Every override is logged.
Rubric-anchored feedback
Rubric-anchored feedback is assembled for the marker. The marker edits before release. The learner receives feedback connected to specific rubric criteria, with the marker's name attached.
Marking capabilities that respect academic judgement.
Written response marking
With AI assistance on, essays and speaking receive draft scores and rubric-anchored explanations. The marker reviews, overrides, signs off.
Spoken response marking
Transcription and criterion-based scoring for spoken responses. Marker reviews at audio level.
Objective scoring
Deterministic scoring for objective items. Configurable partial credit and synonym acceptance.
Rubric-anchored feedback
Feedback connected to rubric criteria. Marker approves before release. Learner sees the marker's name.
Inter-rater visibility
Agreement rates, override rates, drift indicators. Visible to QA.
Audit trail
Every AI output, every override, every sign-off logged. Ready for QA review and candidate appeals.
Marking for the work programmes actually need.
Daily formative feedback
Learners submit a paragraph. AI applies the rubric and prepares rubric-anchored notes. The teacher reviews in batches and releases.
- Teacher reviews in batches
- Connected to learning objectives
A national essay component
Essays across multiple subjects. Your rubric applied consistently across the cohort, borderline cases surfaced first. Markers review, sign off, with statistical moderation on samples and borderlines.
- Standard rubric applied consistently
- Sample and borderline moderation
- Marker sign-off on every grade
Language proficiency speaking exam
Speaking responses transcribed and scored against the framework rubric — pronunciation, fluency, coherence, grammar. Markers review at audio level, adjust where needed, sign off. Per-criterion profile released to the candidate.
- Framework-aligned scoring
- Marker reviews audio
- Per-criterion profile to candidate
Corporate competence verification
Compliance short-answer items scored against the criterion-referenced rubric. Compliance lead reviews borderlines and signs off. Audit trail ready for the regulator. The compliance lead decides who is offered targeted retraining.
- Criterion-referenced rubric
- Compliance sign-off
- Audit trail for regulator
- Retraining decided by the compliance lead
Why this is marking, not auto-grading.
AI is a documented second rater
The marker is still the marker. AI applies your rubric and surfaces evidence; the marker decides.
Override and contest
The marker can override any AI score. The candidate can contest any released grade through your appeals process. The audit trail supports both.
Rubric-anchored, not generic
AI applies your rubric — not a generic model's opinion. Same rubric used in authoring, marking and feedback. Same rubric an examiner would use.
Inter-rater visibility
Agreement rates between AI and markers, between markers, and across cohorts are visible to QA. Drift can be spotted and addressed.
Connected to the lifecycle
Marking flows into Report (analytics) and Certify (verifiable credentials). Same rubric, same audit trail, end to end.
Questions, answered.
See marking that holds up to your QA review.
Bring a sample of marked work. Walk through the workflow with your QA officer. See the audit trail, the override workflow, and the inter-rater data.