Two AI manuscript review platforms with different approaches to research assessment and revision.
Fukurō and Refine both analyze research manuscripts and provide feedback on substantive problems rather than limiting the review to grammar or formatting.
Refine emphasizes technical verification. It examines methodology, logic, argumentation, statistical validity, internal consistency, figures, tables, and related problems. Its current features include supporting documents, comments linked to manuscript passages, interactive chat, feedback management, and revision checking.
Fukurō evaluates research contribution, methodology, experimental design, results, argumentation, structure, citations, limitations, and conclusions. Its AI chat lets researchers discuss specific findings or manuscript passages and ask for possible ways to address identified problems. The Complete plan adds figure and visual evaluation, journal guideline compliance, rebuttal response analysis, and manuscript version comparison.
Both platforms review the research itself. The main differences are the type of analysis they emphasize and the publication stages they support.
Bottom line
Choose Fukurō if you want reviewer-style manuscript assessment, interactive discussion of specific findings, guidance on possible revisions, and dedicated tools for journal requirements, actual reviewer comments, rebuttal responses, and revised manuscripts.
Choose Refine if detailed technical verification is your main priority, particularly for manuscripts containing mathematical reasoning, proofs, statistical analysis, tables, figures, or complex dependencies across sections. Refine also supports supplementary material and provides API and MCP access.
The products overlap substantially. Both provide figure analysis, interactive chat, and revision checking. Fukurō differs most clearly through itsjournal guideline compliance audit,analysis of actual reviewer comments and author responses, and approach of suggesting possible revisions without automatically rewriting the manuscript.
Feature comparison
| Capability | Fukurō | Refine |
|---|---|---|
| Full manuscript review | ✓ | ✓ |
| Research contribution | ✓ | Substantive arguments assessed |
| Methodology | ✓ | ✓ |
| Experimental design | ✓ | ✓ |
| Results and supporting evidence | ✓ | ✓ |
| Logic and argumentation | ✓ | ✓ Strong emphasis |
| Statistical analysis | SAMPL statistical reporting checks | Statistical validity |
| Internal consistency | ✓ | ✓ Strong emphasis |
| Figure and visual evaluation | Complete | ✓ |
| Supporting documents | Not listed as a separate workflow | ✓ Beta |
| Interactive AI chat | ✓ | ✓ Beta |
| Discussion of specific findings | ✓ | ✓ |
| Suggestions for addressing identified problems | ✓ | ✓ |
| Feedback management | Findings with severity levels | ✓ Done / dismissed tracking |
| Revision checking | Complete | ✓ Beta |
| Journal guideline compliance | Complete | No dedicated journal-specific audit publicly listed |
| Rebuttal analysis using actual reviewer comments | Complete | No dedicated workflow publicly listed |
| Manuscript rewriting | Does not automatically rewrite | Chat can assist with rewriting |
| API / MCP | No public workflow listed | ✓ Beta |
| Published document limit | No equivalent published limit | 70,000 words or 50 MB |
| One-time review purchase | — | ✓ |
| Subscription | ✓ | ✓ |
Complete = available with Fukurō Complete.
“No dedicated workflow publicly listed” means that the current public product information does not describe a separate feature for that task. It does not mean that a general review cannot identify a related problem.
Last reviewed: September 2026. Features and pricing may change.
Both platforms review substantive research problems
Neither Fukurō nor Refine is primarily a language checker.
Fukurō evaluates:
- research contribution;
- methodology;
- experimental design;
- results;
- argumentation;
- structure;
- citations;
- limitations;
- conclusions.
Its findings include explanations and severity levels so authors can distinguish major concerns from smaller issues.
Refine similarly focuses on substantive problems. Its documentation describes checks involving logic, methodology, argumentation, statistical validity, internal references, inconsistencies between tables and text, notation errors, and incomplete reasoning.
The relevant difference is therefore not whether either platform examines the research. Both do.
The distinction lies in what they emphasize and how the author can work with the resulting feedback.
Refine emphasizes technical verification
Refine places strong emphasis on checking technical details across the manuscript.
Its public documentation describes a process that follows arguments across sections rather than evaluating each section independently.
Examples of problems it is designed to identify include:
- inconsistencies between text and tables;
- notation errors;
- incomplete proof steps;
- contradictions across sections;
- methodological weaknesses;
- statistical problems;
- logical gaps;
- inconsistencies involving figures.
Refine states that it is particularly effective for manuscripts involving mathematical or logical reasoning. It supports fields including medicine, computer science, law, economics, natural sciences, engineering, mathematics, and statistics.
Refine is therefore particularly relevant when the main concern is finding technical, logical, or internal consistency problems in the manuscript.
Fukurō evaluates the manuscript from a reviewer perspective
Fukurō examines whether the contribution is presented clearly, whether the methods support the conclusions, whether the results support the claims, whether limitations are adequately considered, and whether the argument remains consistent across sections.
An important finding does not need to be a mathematical error or factual contradiction.
It may instead involve:
- a contribution that is not sufficiently supported;
- an experimental design that cannot justify a stated conclusion;
- an unsupported claim;
- a missing methodological explanation;
- a limitation that changes how the results should be interpreted;
- inconsistent descriptions of the contribution;
- an argument that extends beyond the available evidence.
This type of assessment focuses on concerns that may arise when another researcher critically evaluates the paper.
Both platforms provide interactive chat
Fukurō and Refine both allow researchers to discuss review findings rather than treating the report as a static document.
Fukurō
Fukurō's AI chat uses the manuscript and review findings as context.
Researchers can discuss a specific finding or manuscript passage and ask questions such as:
- Why is this considered a problem?
- Which part of the claim lacks support?
- How important is this issue compared with the other findings?
- What additional explanation may be needed?
- What evidence could strengthen the argument?
- What are possible ways to address this problem?
The chat can therefore help the researcher move from understanding a finding to considering possible revisions.
Fukurō does not automatically rewrite the manuscript. It can suggest directions for addressing a problem, but the researcher decides what to change and writes the final text.
Refine
Refine's beta chat can be opened globally or from a specific review comment.
It allows researchers to ask for clarification, question whether feedback applies, explore alternatives, and turn comments into revision steps.
Refine's public chat documentation also lists paragraph rewriting as a possible use of the feature.
The distinction is therefore not whether either platform has chat. Both do.
The difference is that Fukurō explicitly keeps the final manuscript writing with the researcher, while Refine's chat can assist more directly with rewriting text.
Both platforms evaluate figures
Figure analysis is available in both products.
Refine states that it analyzes charts, diagrams, tables, and figures for clarity, accuracy, and consistency with the manuscript.
Fukurō Complete includes figure and visual evaluation.
Figure analysis is therefore not a unique capability of either platform.
Refine can review supporting material
Refine allows users to attach supporting documents with the main manuscript.
Its current beta feature supports materials such as:
- appendices;
- supplementary material;
- datasets;
- protocols;
- robustness tables;
- codebooks;
- supporting documentation.
Refine can consider these materials when evaluating the main paper.
This can be useful when evidence required to understand the manuscript is located outside the main document.
Fukurō does not currently present supporting-document analysis as a separate public feature.
Both platforms support revision checking
Revision checking is available in both products.
Refine's beta Revision Check allows an author to upload a revised manuscript and compare it with the previous Refine review. Earlier comments can be classified as resolved, partially addressed, or still requiring attention. Refine can also identify new problems introduced during revision.
Fukurō Complete provides manuscript version comparison. It identifies additions, deletions, structural changes, citation differences, and whether important revisions appear to have been implemented.
Both platforms can therefore help determine whether problems identified during an earlier review were addressed.
The main difference appears when the revision follows comments from external journal reviewers.
Fukurō analyzes actual reviewer comments and responses
Fukurō Complete includes a dedicated rebuttal response workflow.
Authors provide the reviewer comments and their proposed responses. Fukurō can identify:
- requests that remain unanswered;
- responses that address only part of a concern;
- implicit concerns in reviewer comments;
- statements that need stronger evidence;
- manuscript changes promised in the response but not implemented;
- clarity or tone problems in the response.
This workflow addresses a different question from manuscript review:
Does the response adequately address what the reviewer requested?
Refine's current Revision Check evaluates a revised manuscript against previous Refine feedback.
Its public product information does not currently describe a separate workflow for analyzing external reviewer comments together with the author's rebuttal.
This is one of the clearest differences between the platforms.
Journal guideline compliance
Fukurō Complete includes a journal guideline compliance audit.
This is intended for researchers who have selected a journal and want to check whether the manuscript follows its submission requirements.
Refine describes its reviews as suitable for demanding academic standards and provides services for journals, but its current researcher workflow does not describe a dedicated audit against the submission requirements of a selected journal.
These are different functions.
Refine evaluates the quality and technical consistency of the manuscript.
Fukurō Complete can separately check whether the manuscript follows the requirements of the selected journal.
Refine supports more file formats
Refine currently publishes a standard maximum document size of70,000 words or 50 MB.
Its supported formats include:
- TXT;
- Markdown;
- PDF;
- DOCX;
- ODT;
- RTF;
- HTML;
- EPUB;
- TeX;
- LaTeX.
Fukurō's current manuscript workflow supports PDF, DOCX, and TXT.
Researchers working directly with LaTeX, Markdown, or other formats therefore have more upload options with Refine.
Refine provides Markdown conversion
Refine can convert PDF or Word manuscripts into structured Markdown while preserving equations, figures, and tables.
Equations can be converted into copyable LaTeX, and the resulting Markdown can be downloaded. Review comments are linked to this structured representation of the manuscript.
Fukurō does not currently provide Markdown conversion or export as a public feature.
This may be useful for researchers who prefer to work with a structured text representation of their manuscript.
Refine provides API and MCP access
Refine currently provides beta API and MCP integrations.
Researchers, laboratories, journals, and developers can submit manuscripts programmatically, retrieve structured comments, and integrate the review engine into external tools.
Fukurō does not currently advertise a comparable public API or MCP integration.
This distinction may be less important for individual researchers but relevant for organizations building automated manuscript workflows.
Privacy and retention
Refine publicly describes a zero data retention approach for its review infrastructure. It states that manuscripts are processed for analysis, are not retained afterward, and are not used for model training.
Fukurō uses a different retention model.
Uploaded files and extracted manuscript text are temporarily retained to perform the analysis and are automatically deleted from active systems within three days after the review is created. Structured review feedback can remain available in the account. Fukurō states that manuscript content is not used to train its own models.
Researchers who require immediate deletion after processing may prefer Refine's stated approach.
Researchers working with confidential manuscripts should review the complete privacy policies of either platform before uploading unpublished work.
Pricing
The products use different payment models.
Fukurō
Standard
$24/month
$240/year
- 25 credits per month
- Full multi-agent manuscript review
- Statistical reporting checks using SAMPL
- Grammar and writing checks
- Severity-rated findings with PDF report
- Email support
Complete
$49/month
$490/year
Everything in Standard, plus:
- Figure and visual evaluation
- Journal guideline compliance audit
- Rebuttal response analysis
- Manuscript version comparison
- Priority support
Refine
First document
Free
No credit card required
Full Review
One-time packages. Taxes are calculated at checkout.
- 1 Full Review
- $59.99
- 3 Full Reviews
- $143.99
- 10 Full Reviews
- $359.99
Fukurō currently requires a subscription.
A review does not have a fixed credit cost. Fukurō displays the estimated credit requirement before the review begins.
Refine currently offers one-time Full Review packages as well as subscriptions. It also advertises a free first document with no credit card required.
The products therefore serve different purchasing preferences.
Refine allows researchers to purchase a specific number of reviews.
Fukurō uses recurring credits and separates Standard and Complete based on the additional publication features included.
Which one fits your workflow?
Choose Fukurō if you want
- evaluation of research contribution;
- methodology and experimental design analysis;
- assessment of results and argumentation;
- citation and limitation checks;
- severity-rated findings;
- statistical reporting checks using SAMPL;
- AI chat for discussing specific manuscript passages and findings;
- explanations of why a finding matters;
- possible ways to address identified problems;
- figure and visual evaluation;
- journal guideline compliance;
- analysis of actual reviewer comments and author responses;
- manuscript version comparison;
- revision guidance while retaining control of the final manuscript text.
Refine may fit better if you want
- intensive technical verification;
- detailed checking of mathematical or logical reasoning;
- statistical validity analysis;
- checks for inconsistencies across tables, figures, notation, and text;
- supporting documents included in the review;
- comments linked directly to manuscript passages;
- interactive chat with rewriting assistance;
- revision checking against previous Refine feedback;
- broader file format support;
- Markdown conversion;
- API or MCP integration;
- zero retention processing;
- one-time review purchases.
Frequently asked questions
Does Fukurō have an AI chat?
Yes.
Fukurō's AI chat uses the manuscript and review findings as context.
Researchers can discuss a specific finding or part of the paper, ask why an issue matters, request clarification, and explore possible ways to address the problem.
Can Fukurō suggest how to address a problem?
Yes.
Researchers can use the AI chat to ask for possible approaches to an identified issue.
For example, the chat may indicate that a claim needs stronger evidence, a methodological choice needs further explanation, or a conclusion needs to be narrowed.
The researcher decides whether and how to apply those suggestions.
Does Fukurō rewrite the manuscript?
No.
Fukurō can explain findings and suggest possible ways to address them, but it does not automatically rewrite the manuscript for the author.
Is Refine only for economics or theoretical papers?
No.
Refine supports several disciplines, including medicine, computer science, law, economics, natural sciences, engineering, mathematics, and statistics.
Its documentation states that it is particularly effective for manuscripts involving mathematical or logical reasoning.
Is Refine only a writing editor?
No.
Refine focuses strongly on substantive research issues.
Its review examines logic, methodology, argumentation, statistics, internal references, tables, figures, and other technical problems.
Is Fukurō only a grammar or structure checker?
No.
Fukurō evaluates research contribution, methodology, experimental design, results, argumentation, structure, citations, limitations, and conclusions.
Grammar and writing checks support the manuscript review rather than define it.
Does Refine evaluate figures?
Yes.
Refine analyzes charts, diagrams, tables, and figures for clarity, accuracy, and consistency with the manuscript.
Fukurō Complete also includes figure and visual evaluation.
Does Refine compare manuscript revisions?
Yes.
Its beta Revision Check compares a revised manuscript against the previous Refine review and identifies which comments were addressed, partially addressed, or remain unresolved.
Fukurō Complete also provides manuscript version comparison.
Which platform is more useful after receiving journal reviewer comments?
Fukurō provides a dedicated workflow for this situation.
Its rebuttal analysis examines actual reviewer comments together with the author's responses.
Refine currently provides revision checking against previous Refine feedback, but its public documentation does not describe an equivalent workflow for analyzing an actual journal rebuttal.
Does Refine rewrite manuscript text?
Refine primarily presents itself as a review system, but its beta chat documentation lists paragraph rewriting as one possible use.
Fukurō takes a different approach. Its chat can explain a problem and suggest possible revision directions, but it does not automatically rewrite the manuscript.
Which platform supports more file types?
Refine.
It currently supports TXT, Markdown, PDF, DOCX, ODT, RTF, HTML, EPUB, TeX, and LaTeX.
Fukurō currently supports PDF, DOCX, and TXT.
Which platform provides better peer review?
There is not enough direct public evidence to claim that either platform consistently provides better peer review.
A reliable comparison would require submitting the same manuscripts to both systems and evaluating their findings using the same criteria.
The clearer distinction is their current emphasis.
Refine focuses strongly on technical verification and detecting logical, statistical, and internal consistency problems.
Fukurō combines manuscript assessment with interactive discussion of findings, revision guidance, journal compliance, analysis of actual reviewer responses, and manuscript comparison.
Technical verification or support through peer review?
Refine is well suited to researchers whose main concern is detecting technical, logical, statistical, or internal consistency problems.
Fukurō is designed for researchers who want to identify manuscript weaknesses, discuss individual findings, explore possible ways to address them, and continue through journal submission and revision.
Review the manuscript. Understand the findings. Address reviewer comments. Verify the revision.