PaperMentor: A Human-Centered Multi-Agent Writing Tutor for AI Research Papers on Overleaf

2026-06-07

AI Summary

This work addresses the longstanding challenge that early-career researchers often lack access to high-quality, actionable feedback on academic writing, as existing AI tools typically offer only grammatical corrections or generic evaluations. To bridge this gap, the authors propose a human-centered, multi-agent writing coaching system natively integrated into Overleaf, which delivers fine-grained, context-aware revision suggestions directly as inline comments while preserving full author control. The system combines an expert-curated knowledge base—constructed by experienced researchers—with twelve specialized agents, each targeting distinct dimensions of scholarly writing, and leverages natural language processing for seamless integration. User studies demonstrate that 90.6% of the generated comments are rated as actionable and 67.5% as effective, significantly outperforming a GPT-5.2 baseline without the expert skill repository. The implementation is publicly released.

Abstract

Expert writing feedback from experienced researchers is critical for early-career scholars to improve their manuscripts, yet high-quality feedback often remains scarce because reviewing research papers is labor-intensive. Emerging AI-powered writing assistants largely focus on grammar fixes or simulating peer review with final scores, yet they fall short of providing concrete, actionable suggestions that help students improve their papers during drafting. We present PaperMentor, a human-centered writing assistant system that delivers actionable suggestions as Overleaf-native inline comments while leaving the actual writing entirely to human authors. PaperMentor integrates an expert skill library carefully curated from established researchers' writing advice with 12 specialized agents covering different aspects of paper writing, such as formatting compliance, phrasing accuracy, and terminology consistency. In a user study (n=14), 90.6% of the generated comments were rated actionable and 67.5% were rated valid, significantly outperforming a GPT-5.2 baseline uswithout the skill library. We release PaperMentor as open source for public use. Our code is publicly available under the AGPL-3.0 license at https://github.com/jiarui-liu/overleaf

Problem

• writing feedback
• research papers
• actionable suggestions
• AI writing assistant
• early-career scholars

Innovation

• multi-agent system
• expert skill library
• actionable feedback
• AI writing assistant
• Overleaf integration