← peerfect.co

Frequently asked questions

Will my manuscript be used to train AI models?

No. Your manuscript text is used to compute your review and recommendations, and stored on our infrastructure (EU). It is never used to train foundation models, never sold, and never shared with publishers.

Which languages are supported?

The full review pipeline runs on Turkish and English manuscripts. Other languages route through the SPECTER2 path (English-aligned embeddings) and produce reasonable but not first-class output. Bilingual journal recommendations work for Turkish manuscripts via BGE-M3 embeddings on a separate index.

What disciplines does Peerfect cover?

The journal corpus covers ~120K international journals (OpenAlex-derived) plus ~2,500 Turkish journals from DergiPark. The audit pipeline is discipline-agnostic at the agent level; recommendations weight discipline-specific signals (top topics, subfields, JCR category) when available.

Is the beta free?

Yes — the closed beta runs free, with the only gate being an invite from the Peerfect team or Esra's network. Pricing is finalized after the beta concludes; we will publish a transparent academic-tier price before turning anything off.

Where is the data hosted?

Postgres, Redis, MinIO, and the embedding service run on a single Hetzner CCX33 node in Falkenstein, Germany (EU). Static assets are served from Cloudflare's edge. Nightly Postgres dumps mirror to AWS S3 (Glacier IR, eu-central-1).

How do I delete my data?

Delete a manuscript from the Inbox to remove its blob within 24 hours; backups roll off in 30 days. Account deletion is in Settings; it cascades to all related manuscripts, runs, and review reports.

Can I run the skills locally?

Yes. The peerfect-skills package is Apache-2.0 licensed; you can run novelty / citation / journal_match / field_classify on your own machine without an account.