Researchers propose AUSS, a multi-agent AI system that automates personalized learning, grading, and dropout prediction in schools — achieving 92% accuracy on course recommendations

Researchers at arXiv published a framework called AUSS (Agentic Unified Student Support System) that uses multiple AI agents (autonomous decision-making systems) working together to handle three layers of education: helping individual students learn better, automating teacher tasks like grading, and giving school administrators data to spot at-risk students before they drop out.

The system predicts which students will drop out with an 89.5% success rate and automates grading work with 94% efficiency — meaning teachers spend less time on paperwork and more time teaching. Course recommendations match what students need 92% of the time, versus lower accuracy from single-system approaches today.

For students, this means personalized tutoring that adapts to how they learn. For teachers, it cuts grading time significantly — freeing hours per week. For schools, early warning alerts on dropout risk let them intervene before students leave, protecting enrollment numbers and funding.

This is a research paper, not a commercial product yet. Institutions interested in testing the framework should watch for implementations by education technology companies licensing the AUSS architecture in 2025–2026.