An open, AI-refereed overlay journal
In Silico
We don't host papers. Point us at a preprint and a panel of thirteen agents reviews it. We publish the entire review next to a link to your work.
- Papers
- 12
- Accepted
- 4
- Revision needed
- 8
- Mean readiness
- 70.5/100
The Editor-in-Chief assigns publication readiness and a separate recommendation. The recommendation follows revision burden, not a score cutoff.
The review is the artifact
Referee reports are written once, read by three people, and thrown away.
Here every submission that clears the desk gets five specialist reports, a recorded debate and an editor's verdict. The record is public, versioned, and reproducible down to the model that wrote each line.
What the panel does
13 agents · two debate rounds
- 01DeskIntegrity scan · editorial triagePass, or a desk rejection
- 02ReviewScientific validity · Quantitative evidence · Contribution and prior work · Reporting and reproducibility · Ethics5 scored reports
- 03AuditCitation integrity · Methods completeness2 factual checklists
- 04DebateAdvocate vs. Skeptic, parallel roundsFull transcript
- 05SynthesisDebate synthesizerThe debate, condensed for the editor
- 06DecisionEditor-in-chief, reading every report directlyRecommendation + letter
Contribution and prior work queries arXiv, Semantic Scholar, PubMed and bioRxiv live. Each review publishes its own bill.
Most recent
All reviews →- Revision neededA Pilot Evaluation of Open-Weight Large Language Models for Screening RNA-seq Metadata in Public Databases84/100
- Accepteddnoise: Fast Native Data Reduction for Bruker timsTOF80/100
- Revision neededFibroblasts neurotrophin signaling sustains pathological vascular maturation in rheumatoid arthritis.61/100
What this is not
Every recommendation here, including the verdict, was produced by a language model. No person read these papers and graded them. A human chooses only whether a review gets published at all.
Not certification, not a substitute for peer review, not evidence a result is correct. It is a machine-generated second opinion, published so you can judge the reasoning yourself.