Publishing a research manuscript is a routine part of scientific work, yet it often consumes time, creates stress, and yields uncertain outcomes. Recent advances in large language model (LLM)‑driven agentic AI have shown promise across many research tasks, prompting us to ask whether such AI can assist researchers in predicting the eventual journal for a manuscript. We introduce PASS (Publication‑oriented Agentic Scientific System), which understands a manuscript within its domain‑specific literature context and returns a ranked list of suitable journals. PASS reconstructs the manuscript's local scientific neighborhood, traces its topic trajectory, and reasons over the field’s journal space to position the work. Evaluation was performed on a leakage‑audited benchmark covering 16 biomedical sub‑fields and over 2,000 preprints. PASS achieved a Top‑1 accuracy of 50.3% and Top‑5 accuracy of 86.1%, surpassing state‑of‑the‑art LLM baselines and existing journal‑selection tools. Quality scores produced by PASS—such as impact potential and novelty—correlated strongly with independent measures of publication outcome. Analysis revealed that the literature retrieval module contributed the most to performance, especially for positioning manuscripts relative to nearby work; PASS also retained near‑full performance using only the abstract, whereas LLM baselines required the full text. An independent human evaluation showed strong researcher agreement with PASS's manuscript understanding and recommendation rationale. PASS has been released as a public platform (https://ratemypaper.ai/) for the research community.
Review: PASS demonstrates the practical value of agentic AI in the scholarly publishing pipeline, with its literature‑driven positioning offering reliable journal‑selection guidance for authors.