Automate grading whilst maintaining academic integrity

Setting the new benchmark in AI grading for education

Our goal is to automate the entire marking process, including scoring and written feedback for all subject areas. Early tests and simulations show a reduction in marker involvement of 70-90% using our bespoke-trained AI models, while maintaining a >90% agreement rate compared to human markers.

This “marker-in-the-loop” approach focuses on aiding existing markers with pre-filled feedback and scores. In the long run, the remaining human marker involvement will be utilised for benchmarking, spot checking, and borderline cases.

Applicable in high-stakes testing for higher education and practice environments, we believe AI grading can play a significant role in reshaping the educational landscape. This movement can lead to more personalised feedback, support, and ultimately a new learning experience.

With a reduction in the marking workload, more teaching resources can be allocated to support students and increase learning efficiency.