Gradr
Gradr is a Swedish platform that provides AI-assisted formative feedback and secure digital exams for schools. The AI generates feedback suggestions based on each teacher's own instructions and grading criteria, which the teacher then reviews, edits, and approves before anything reaches students. The platform supports whole-class grading, locked exam mode, and EU-based data hosting with no AI model training on user data.
- Artificial Intelligence
- Education Technology
- Software Only
Funding
Founders
Product
Problem
Teachers spend significant time writing individual formative feedback for each student, which limits how often and how detailed that feedback can be. This results in students receiving generic or infrequent comments that do not help them improve their learning effectively.
Solution
Gradr is a platform for AI-assisted formative feedback and secure digital exams, built for schools. The AI proposes feedback for each student based on the teacher's own instructions, assessment criteria, and the curriculum, and the teacher reviews, edits, and approves every comment before it is published. Teachers can upload whole-class responses as PDF or DOCX files or run exams directly in Gradr's locked exam mode. The platform also provides a student interface where learners can write, submit, receive feedback, and revise their work in one place, with individual accommodations like text-to-speech and extended time available.
Target Audience
Primary users are teachers and students in Swedish primary and secondary schools, with the platform currently used by over 500 teachers across more than 100 schools.
Features
- AI feedback generation grounded in teacher-defined criteria, tone, and curriculum alignment, with the AI asking for clarification when uncertain
- Whole-class grading that processes all student responses simultaneously without mixing up submissions
- Built-in exam platform with open or locked mode, deterministic multiple-choice grading against the teacher's answer key, and individual accommodations such as text-to-speech and extended time
- Teacher approval workflow where no comment reaches a student without active review and editing by the teacher
- EU-based hosting at Hetzner in Finland with AES-256 encryption at rest, TLS in transit, and a data processing agreement prohibiting AI model training on user data
- SSO integration and optional mandatory two-factor authentication for access control