Research Summary: Edhivo: An AI-Native Platform for Scalable Assessment and Instant Feedback in Higher Education

Edhivo is an AI-native pedagogical platform developed to address higher education assessment bottlenecks, specifically the faculty burden of test authoring and the operational difficulty of semantically evaluating open-ended responses. Utilizing a vertically integrated infrastructure and a Dual-LLM pipeline, the platform generates context-aware assessments from instructor materials and conducts multi-criteria, rubric-based student evaluations to facilitate rapid feedback cycles.

Why it matters

This development marks an operational shift toward specialized, multi-model AI pipelines designed to solve domain-specific workflow bottlenecks rather than relying on generalized conversational interfaces. For higher education and training organizations, automating both curriculum assessment generation and subjective evaluation directly addresses instructional capacity limitations. It enables scalable, continuous feedback loops without requiring a linear expansion of faculty or grading resources.

Key insights

• Higher education assessment faces dual operational constraints: high cognitive effort in authoring assessments and limited capacity to semantically grade open-ended student work.
• The Edhivo platform utilizes a vertically integrated system that combines content delivery, automated assessment generation, and reporting.
• The underlying Dual-LLM architecture separates duties: the first model creates context-aware multiple-choice and open-ended assessments from course content, while the second performs rubric-based semantic evaluations.
• The platform aims to bridge the student feedback gap by providing near-instantaneous, multi-criteria evaluations of complex submissions.
• The source abstract is truncated, limiting access to empirical validation data and final performance outcomes.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2610.36654