Manusights examples
Inspect the source manuscript and the engine output side by side: same review your paper would get, run on public arXiv and bioRxiv papers so every claim is verifiable.
A desk rejection can cost months in the queue. Every example below is the kind of issue that triggers one, caught before submission.
Full Review · $39
7-dimension scorecard · 14+ findings · target-journal call
A theoretical condensed-matter paper deriving logarithmic decay near an A₃ tricritical singularity from replicated overlap statics. Strong derivation, but sign-convention drift across six equations and a factor-of-four prefactor discrepancy in the main-result chain need a coordinated audit before submission.
View exampleA simulation-based-inference method for epidemic models when the underlying contact network is only partially observed, applied to a real Shark Bay dolphin tuberculosis-skin-disease dataset. The method is genuinely interesting; the manuscript currently has model-definition contradictions and a missing reproducibility package that would draw a desk-reject from PLOS Computational Biology.
View exampleA retrospective quasi-experimental analysis of Germany’s 2015 national minimum wage, using the Structure of Earnings Survey across three waves to estimate effects on hours, minijobs, and wage inequality. The minijob institutional angle is a real contribution; magnitudes do not yet reconcile across sections and the identification package is asked to carry too much interpretive weight for a Labour Economics submission.
View exampleMarker-based CRISPR screens nominate the transcription factor POU2F1 as an activator of DLL3 — the small-cell-lung-cancer antigen behind tarlatamab and other approved therapies. The discovery arc is strong, but the headline "cis-regulatory code" claim is still correlative, an AlphaFold analysis is asked to carry mechanistic weight it cannot, and the ChIP-seq peak calling skipped its own input controls — the kind of issues a Cancer Cell reviewer raises before the science is debated.
View exampleA double-blind, placebo-controlled RCT reporting that a probiotic lowered plasma pTau181 and preserved visuospatial cognition in older adults with early cognitive impairment. There is a real signal here — but the trial is reported as confirmatory without a registration, a prespecified primary endpoint, or multiplicity control; several tables are internally impossible (a total mean that cannot be produced by its own group means); and the title’s "ameliorate tau pathology" claim runs well past what a single plasma marker can show. This is the kind of review that decides a CONSORT desk-reject.
View exampleA timely method for stabilizing asynchronous reinforcement learning of LLMs: it diagnoses training collapse as a variance problem predicted by collapsing effective sample size, and proposes an ESS-scaled learning rate plus a closed-form off-policy baseline. The idea is good and the code is released — but the implemented gradient estimator is under-specified and potentially biased, the "minimum-variance" baseline is derived for the wrong importance weights, and the headline reliability and 2.5× speedup claims rest on single training runs with no seeds or error bars. Exactly what a NeurIPS reviewer pushes on first.
View exampleBuilt for your revision rounds
When you upload a revised version, the engine reviews it in light of your prior round and tells you, per issue, what you addressed, what still shows up, and what’s new — something a one-off AI chat cannot do. Most manuscripts go through two or three rounds; the 3-pack and 5-pack (24–30% off) are built for exactly that.
Submission-Ready Dossier · $99
Cover letter · reviewer simulation · response strategy · everything ready to submit
A public bioRxiv Dossier for a pancreatic cancer manuscript, pressure-testing Nature fit, causal overclaim risk, human-validation gaps, statistics, figures, citations, and reviewer strategy.
View full DossierA public arXiv Dossier for a spatial-transcriptomics methods manuscript, stress-testing Nature Methods fit, reproducibility gaps, benchmarking claims, figure readiness, code/data availability, and submission packaging.
View full DossierA public arXiv Dossier for a climate-attribution statistics manuscript, stress-testing Nature Climate Change fit, policy-facing overclaim risk, simulation validity, uncertainty reporting, reproducibility, and reviewer strategy.
View full DossierFrom researchers
Quotes used with permission, taken from public posts, or drawn from researchers' own written feedback to us, from authors who ran the review on a real manuscript headed for a real journal, across fields.
“The feedback I got from Manusights was both candid and correct about the strengths and limitations of my manuscript. I've seen some of this feedback before, the hard way. Seeing it earlier would have been useful.”
Bo Cowgill
Assistant Professor, Economics
Columbia Business School
“Honestly, I really like this tool, perhaps a bit too much. Manusights has been extremely helpful in identifying realistic target journals. From my experience, it's actually the only AI assistant that gives feedback on the full manuscript rather than only the abstract. With ChatGPT there is often a kind of confirmation bias, which is much less present in Manusights. It helped me substantially restructure my paper after a peer-review rejection. It's now under review at AJAE.”
Anaëlle Denieul-Babrot
Recent PhD, Economics
INRAE, France
“Manusights is invaluable to researchers and clinicians looking for the best publication venue for their work. The AI analysis is extremely helpful and on target. This is exactly where I see the real value of AI models to date.”
Deborah E. Sewitch, Ph.D.
Sleep medicine clinician-researcher
Diplomate, American Board of Sleep Medicine
“The Manusights evaluation of my draft was like having your own peer reviewer on demand. The positives were appreciated, and the negatives were spot on, giving me a 'why did I not think of that' kind of moment. The recommendations are actionable and helpful.”
G.T.B.
Practising clinician, Singapore
BMJ Case Reports submission
“Standard peer review doesn't do this. It filters. It withholds. It tells you whether you passed a gate, not what your work actually needs. What Manusights gave back was specific, claim-level, actionable engagement with the actual science.”
Hector Damian Cirino
Independent Researcher, Atmospheric Science
Harmora Framework
“I've tried a few AI review tools, but Manusights stands out, almost like having a tutor. It foresees what editors and reviewers will object to and drafts responses with exact changes needed. It caught papers I'd missed, all verified, saving another round of searching. Every round builds on the last, and just knowing where to submit saved me months.”
Lora Chen
Researcher, Biomedical Research
University of Hong Kong
“I was very impressed with Manusights' report. It was as if a human editor had reviewed and replied. It was valuable, and I incorporated several recommendations into the manuscript before submitting.”
Terry Grimmond
Consultant Microbiologist
Grimmond & Associates
“Helpful in pointing out particular reviewer objections and concerns, and confirmed that the journal is appropriate for submission.”
Freddi Segal-Gidan, PA, PhD
Director, USC/Rancho Alzheimer's Center
Keck School of Medicine of USC
“Yes, very helpful, and the tool is impressive. I came back with a few suggestions for future versions, not complaints.”
Luis Lemus
Associate Professor of Cognitive Neuroscience
IFC, UNAM
“So we had this censoring issue in our survival analysis that nobody caught. And I mean nobody, two of my co-authors have biostatistics backgrounds. The report flagged it in like paragraph three. We fixed it before sending to PLOS Comp Bio and not a single reviewer brought up the stats.”
Sarah L.
Postdoctoral Fellow, Computational Biology
Biostatistics group, R1 university (US)
“The feedback on our discussion was specific. It said we were treating a correlation as mechanism, and pointed to the exact paragraph. Changed the framing, submitted to JEM, got major revision instead of reject.”
James W.
Associate Professor, Immunology
School of Medicine (US)
“I'm not a native English speaker and I'd already paid for language editing separately. This was completely different. It wasn't about grammar at all. The report basically said my introduction didn't match what Nature Communications actually publishes in my subfield. Very specific, not generic advice.”
Wei C.
Assistant Professor, Systems Biology
National university (East Asia)
“It said Figure 3 was unreadable without two paragraphs of context from the results. We redesigned it before submitting to eLife, and a reviewer in round two specifically said the figures were well-constructed.”
Emma B.
PhD Candidate, Neuroscience
University medical center (Northern Europe)
“The advice sharpened the paper before I sent it out. It's now submitted to Monthly Notices of the Royal Astronomical Society.”
J.
Astrophysicist
Ivy League university (US)
Most testimonials are anonymized to protect client confidentiality. Named quotes are used with permission, quoted from the author's own public posts (lightly edited for clarity, meaning preserved), or drawn from the researcher's own written feedback to us.
You have seen the output. Start your Full Review ($39) or Submission-Ready Dossier ($99) now and upload your manuscript right after checkout.
Working with a co-author?
Send a researcher who is new to Manusights to /ai-review; once their first review is paid, we credit one free Full Review to each of you.
Trained by current peer reviewers
The Manusights engine was built with 35+ active reviewers, including current Nature, Cell, and Science reviewers, who trained it on their real peer reviews.