Many journals, study sections, and funders prohibit AI use in peer review. Some require disclosure of any AI assistance. Some allow it for limited purposes. Their rules are the rules. Do not break them.
If you are bound by a no-AI policy and you still want to use this tool, use it as a guide only, not a decision-maker:
Read the panel reports for what they catch that you might have missed, then write your own review in your own words from your own reading.
Do not paste the panel's text into a real reviewer report or submission portal.
Do not let the panel's decision substitute for your professional judgment.
If your journal or study section requires disclosure of AI assistance of any kind, disclose it.
Transparency is the only way to use AI correctly. If you cannot disclose your use of this tool to the journal, study section, mentor, or institution you are accountable to, do not use it for that work.
This tool is intended for: pre-submission self-review of your own manuscripts; rehearsal for students learning the peer-review process; helping authors anticipate reviewer concerns. It is not intended for: generating peer reviews you will submit to a journal as if you wrote them; reviewing grants under a no-AI policy; any context where AI assistance is prohibited or requires disclosure you cannot make.
Office Hours · Reviewer Panel
AI-simulated reviewer panel.
Upload your manuscript. Three reviewers (each an AI persona modeled on a distinct academic lens) will read it independently and write their own reports in their own voice. Methodologist, Domain Expert, Skeptic, Theorist, Stats Hardliner, Outside Reader, Interdisciplinary Bridge, Supportive Editor. Editor decision computed from the panel. Built as a pre-submission rigor check, not a replacement for real peer review.
What this is: Each reviewer is an AI persona running on Anthropic’s Claude API. The personas are fictional academic archetypes; the reviews they write are AI-generated. This tool is for pre-submission rigor and rehearsal; journal peer review remains the authoritative process. Anthropic does not train on or retain API content. We do not log or store anything on our servers. Use a redacted draft if it includes patient identifiers or sensitive collaborators.
Drop your paper PDF here, or click to upload
Under 4 MB. Single PDF, up to ~100 pages. Anthropic reads the document directly.
file
Pick your reviewers
Three are pre-selected based on your paper type. Click a portrait to swap in or out. Pick 2 to 5 total.
Reviewer panel report
Editor decision is computed from the panel. Each reviewer's report is editable in place.