Agentic AI Peer-Review for Research Manuscripts

Reviews are generated by artificial intelligence and errors may be encountered. Since the system is grounded in arXiv data, higher accuracy is typically achieved in fields such as AI where recent research is published freely. Variations in accuracy may be observed in other disciplines.

Only English-language papers are supported at this time. While various disciplines can be processed, the highest level of performance is achieved on topics extensively covered in open-access repositories like arXiv.

This tool was designed for researchers to obtain feedback on their own work. The use of this platform in any manner that violates established peer review policies is strongly discouraged for conference reviewers.

The academics-grade Agentic Reviewer ingests a paper PDF (optionally with a target venue), converts it into structured Markdown, and verifies that the manuscript meets core academic document criteria. To ground evaluation in current scholarship, it generates diverse search queries spanning benchmarks, competing methods, and adjacent techniques, then runs deep retrieval workflows and links evidence to arXiv records (arxiv.org). The system ranks relevance, determines whether abstract-level evidence is sufficient or full-text synthesis is required, and when needed processes full papers into focused technical summaries. Finally, it synthesizes manuscript evidence with curated related-work findings into a structured, academically rigorous review template.