Marking doesn't scale Inconsistency does
A lecturer marking 300 essays over a weekend doesn't give the same quality of feedback on essay 280 as they did on essay 1. That's not a failure of the lecturer, it's a failure of the system. Inconsistent grading creates appeals, erodes student trust, and puts institutions at risk. AI grading doesn't get tired.
Marking consistency over a batch of 300
1–60
61–120
121–180
181–240
241–300
Marking quality across a typical batch
Within 5%
of human marking, every
submission
97%
of grades unchanged after review
Marking quality drops at scale
Essay 280 gets less attention than essay 1. Every time.
Inconsistency creates appeals
When two students with the same work get different grades, institutions face challenges.
Feedback arrives too late
By the time students receive their grade, the learning moment has passed.
Consistent marking
Eduface applies the same rubric criteria with the same weight every time.
Assessment Results
6.6 / 10
Argument structure
Weight: 20%
8 / 10
Use of evidence
Weight: 20%
7 / 10
Academic writing
Weight: 20%
5 / 10
Critical analysis
Weight: 20%
7 / 10
Referencing
Weight: 20%
6 / 10
Within 5%
of human marking, every submission
Calibrated to your rubric, not a generic model
97%
of grades unchanged after review
Lecturers approve, they don't re-do
100%
of decisions are auditable
Every grading decision is explainable.
Grading your way
Four formats, one rubric. Configure once, reuse across every cohort.
Points-Based Grading
1 to 10 points per criterion
7 / 10
Descriptive Levels
Unsatisfactory, Satisfactory, Good, or Great
Satisfactory
Pass / Fail Grading
Simple binary outcome per criterion
Pass
Descriptive Extended
Six levels from Low Fail to Outstanding
Outstanding
Procurement-ready out of the box
Every regulatory requirement your institution faces is already addressed
EU AI Act
Article 13 Transparent & explainable
Eduface uses its own model, not a third-party black
box. Every grade is explainable and auditable, meeting
Article 13 transparency requirements directly.
Own proprietary model, fully auditable
Every grading decision is explainable
No reliance on opaque third-party APIs
Compliant with high-risk AI system requirements
UK GDPR
Your data, under your control
Student data is processed in compliance with UK
GDPR. No personal data is used to train models. A Data
Processing Agreement is available on request.
Student data stays within your region
Student data stays within your region
DPA available on request
ICO-aligned data handling practices