How it works

A second opinion,
never a verdict.

Upload your manuscript. Get back a structured review that reads like it came from a careful reviewer at your target journal — findings tagged by severity, paired with the section they came from. how to fix it.

We surface what a reviewer would surface. The decision to accept, revise, or set aside is yours.

Try free reviewSee a sample review →
N Engl J MedDraft 3 · Under review
Effect of a Structured Remote Patient Monitoring Program on 180-Day Hospital Readmission in Heart Failure
Ramirez DA, Patel SK, Okafor NE, et al.
§ Abstract
Methods A total of 200 patients were enrolled at a single academic center.
Results The primary composite of readmission or ED visit was reduced (HR 0.67; P=0.048); mortality was not significant in ITT.
Conclusions RPM supports routine implementation as a standard component of post-discharge HF management.
M1
Major
ITT vs. mITT analysis description is internally inconsistent.
M2
Minor
Mortality CI upper bound essentially touches 1.0.
M5
Major
‘Routine implementation’ conclusion not supported by data.
Examples use a synthetic manuscript created for demonstration. No real patient data, author work, or published research is shown.
The pipeline

Upload to insight, in five stages.

01

Choose Your Depth

Pick the mode that matches your situation — from a focused journal-club read to a co-author revision pass with tracked changes in your draft.

EssentialsPeer ReviewAuthor Review
02

Upload Your Manuscript

.docx or .pdf — we extract the full text, detect formatting metadata, and identify your target journal.

03

Manuscript Analysis

Our detection engine identifies your study type, matches the correct reporting guideline, and selects discipline-specific evaluation criteria.

04

Multi-Stage Expert Review

Not a single prompt. A structured multi-stage analysis: article and journal-specific deep review, true reference verification, compliance audit, and quality validation.

Unlike chatbot wrappers, PeerReviewAI uses actual reporting guidelines, expert article-specific methodology resources, and journal-specific author instructions — not just a prompt and an LLM.
05

Export & Share

Download your review as a premium PDF, generate tracked changes in your .docx, and email directly to co-authors.

What's in the review

Eight dimensions, one careful pass.

Platform capability. Each tier above turns up the volume on how deeply we go.
Scientific rigor
01
Conclusions vs. Evidence
Claims checked against what the data actually shows. Overstatements and leaps flagged.
Found:
“Supports routine implementation of RPM” — 5 vs 12 deaths; CI 0.16–0.98.
02
Literature Context
How the work sits in the existing evidence base — consistent, novel, contradictory, or already-answered.
Found:
BEAT-HF (Ong, 2016) and Tele-HF (Chaudhry, 2010) found no benefit — not addressed.
03
Statistical Claims Audit
Power, effect sizes, significance language, multiple-comparison handling, and reported vs. required sample sizes.
Found:
5 secondary outcomes reported; methods state no multiplicity adjustment.
04
Practical Implications
How the findings translate — or don't — to the real world, the clinic, the bench.
Found:
Single-center, LVEF ≤40% only — conclusions generalize to all HF patients.
References & compliance
05
Reference Verification
Citations checked — that they exist, say what the manuscript claims they say, and are correctly attributed.
Found:
Ref 9 (Harrison, 2022) cannot be located in PubMed — possibly fabricated.
06
Missing References
Claims and methods that would ordinarily cite prior work, but don't — surfaced with candidate references to consider.
Found:
TIM-HF2 (Koehler, 2018) is the most directly comparable trial — not cited.
07
Reporting Checklist Compliance
Whether the manuscript reports the elements expected for its study type, section by section.
Found:
Missing: allocation concealment mechanism, blinding of outcome adjudicators.
08
Journal-Specific Compliance
Formatting and submission requirements for your target journal — word counts, structure, figure specs, disclosures.
Found:
Abstract 314 words — NEJM limit is 250. Structured format required.
Why this is different

Not a Chatbot Wrapper.
A structured, methodological, multi-stage pipeline.

Reporting guidelines, journal-specific author instructions, and article-specific methodology resources — used to confidentially evaluate your manuscript. Not a single prompt with a generic response.

Grounded, not hallucinated.
Findings are checked against established reporting guidelines, methodology standards, and verified references — not generated from memory.
Consistent, not random.
Structured evaluation criteria produce reproducible assessments. The same manuscript gets the same rigor on the next pass.
Protected by Anthropic’s Zero Data Retention.
Your manuscript is never saved, logged, or used for training.
See it in motion

Author mode, watch it work.

Real-time AI review generation — watch the review write itself, line by line.
Two short demos · sound on
Demo 01 · Author Review
Video coming soon
Demo 01 · Author Review
A structured review, generated live.
Upload your manuscript. Watch the review scroll in — findings tagged by severity, anchored to the section, paired with a fix.
Demo 02 · Tracked Changes
Video coming soon
Demo 02 · Tracked Changes
Tracked changes, applied in place.
Suggestions land in your .docx as tracked changes — insertions, deletions, comments. Accept, reject, revise. You stay in control.
Real captures · unedited · typical manuscript length
Special section · ESL authors

For authors writing in a second language.

A dedicated review section addresses idiomatic phrasing, tone calibration, and technical vocabulary consistency — without flattening your voice or style.

  • ›Clarity rewrites — same meaning, plainer English
  • ›Tone calibration for target-journal register
  • ›Technical vocabulary consistency across sections
  • ›Idiomatic phrasing — where it helps, where to avoid
Original
"The obtained results have been demonstrated that the proposed approach is able to achieve the significant improvement in the patient outcomes."
↓
Suggested
"These results show that the proposed approach significantly improves patient outcomes."
24 → 13 words · voice preserved · same meaning
JOURNAL CLUB · STAY CURRENT

Research intelligence, delivered to your inbox.

PeerReviewAI monitors PubMed weekly and synthesizes what matters in your field — so you spend less time searching and more time reading what counts.You choose your topics and preferred journals in your profile — every edition is built around what matters to you.
Monthly Digest
FREE
Monthly Digest

A curated synthesis of recent research:

  • Choose up to 2 topics and 2 journals
  • Multi-paper synthesis across your chosen topics
  • Evidence direction and clinical relevance for each finding
  • Bottom-line takeaway you can act on
  • Delivered the first Sunday of every month
Weekly Edition
$19.99/MO
Weekly Edition

Seven content formats, rotating weekly:

  • Choose up to 5 topics and 5 journals
  • Latest Research — multi-paper synthesis of what just published
  • Questions — AI identifies a question the literature can answer, then finds the papers that answer it
  • Landmark Papers — deep analysis of foundational work
  • Historical Narrative — a serialized chapter-by-chapter history of your field, delivered one chapter per week
  • Methodology Deep Dives — how the science gets done (trial design, statistical methods, systematic review technique)
  • Clinical Scenarios — case-based reasoning that connects evidence to practice

Plus a "What's New" brief in every edition — no repeated conclusions, no recycled content.

Start premium
Ready when you are

Upload a manuscript.
Get a real review back.

FAQ

Questions, answered.

Don't see yours? Email us — we read every one.