Manuscript content is not used for training
Uploaded manuscripts are processed to produce the requested review output. Manuscript content is not sold, shared for advertising, or used to train Manusights models.
Trust Center
Manuscript privacy and institutional review in one place.
Manusights handles unpublished manuscripts as confidential research assets. This page collects the operating boundaries, vendor review materials, and policy links researchers and institutions usually need before using a manuscript-review service.
Access is purpose-limited
Automated review workflows use service access for ingestion, processing, report delivery, support, and reliability. Human review paths require a separate expert-review engagement.
Retention is bounded by the product workflow
The system keeps operational records needed for report delivery, billing, support, abuse prevention, and auditability. We avoid keeping manuscript content beyond what the product workflow requires.
Institutions can request due-diligence materials
Labs, departments, and libraries can request data handling details, vendor review answers, billing documentation, and a DPA discussion before a paid pilot.
Review standard
Credibility comes from reviewer-calibrated language, not a hidden human review claim.
Manusights is differentiated from generic LLM feedback because the product is organized around reviewer objections and submission risk. The automated path stays clearly separated from Expert Review so authors know exactly what they are buying.
35+ reviewer calibration
The automated review standard was shaped by early work with 35+ CNS-experienced reviewers and senior scientists. Their language informed how Manusights describes unsupported claims, weak controls, novelty gaps, journal-fit risk, and likely reviewer objections.
Automated path boundary
Automated Full Reviews and Dossiers do not use a live human reader on this path. Manusights generates the report through the product workflow; optional Expert Review is separately scoped when a human scientist reviews the manuscript.
Honest outcome limits
Reviewer-calibrated diagnostics can identify submission risk before you submit, but they do not replace journal peer review, guarantee acceptance, or predict every reviewer reaction.
Subprocessors
Core vendors used to operate Manusights.
This list is intentionally practical. It explains the systems that may receive account, operational, billing, email, hosting, or review-task data while Manusights provides the service.
Vendor
Purpose
Data involved
Anthropic
AI manuscript analysis and review generation
Manuscript text and task context required for the review
OpenAI
Structured-output fallback and selected review tasks
Task context required for the review when that path is used
Supabase
Authentication, durable job state, billing metadata, and app records
Account, job, attribution, and operational metadata
Stripe
Payment processing, receipts, refunds, and billing records
Payment and customer billing data
Resend
Transactional and lifecycle email delivery
Email address, delivery metadata, and message content
Vercel
Website and application hosting
Standard hosting, request, and deployment logs
Due diligence
Documents for researchers, labs, and institutions.
Manuscript data handling brief
Detailed confidentiality boundaries, retention language, access-control expectations, and non-claims.
Open documentPrivacy Policy
Customer-facing privacy terms, retention language, third-party services, and contact path.
Open documentSecurity overview
Encryption, infrastructure, access-control, and compliance overview.
Open documentTerms
Commercial terms for use of the Manusights website and review services.
Open documentImportant boundaries
What Manusights does not claim.
- Manusights does not replace journal peer review.
- Manusights does not guarantee acceptance, reviewer sentiment, or editorial outcome.
- Manusights does not make confidentiality risk zero. The goal is clear boundaries, limited access, and practical controls.
- Human expert review, when purchased, has a separate access model from automated Full Reviews.