First ethical AI peer reviewer

A paper is an argument.
Readmy maps the structure.

Readmy is an editorial instrument, not a verdict machine. It marks where language departs from the role a section performs, then hands that map to the people whose judgement the process depends on.

18
unique markers
0
accept or reject calls
1
pass over legible text
ms-4471 / readmy-indexLive map
Introduction
Low
Methods
Low
Results
Medium
Discussion
High

The workbench

Every marker in view, every marker traceable.

Select a marker in the manuscript or in the report panel. The traceability drawer names the attribute that produced it, the role of the section it sits in, and what was compared.

readmy.org/example/manuscript-preview

Document overview

ReviewAttentionFocus

Document copies

The core premise

A research paper is an argumentative structure.

Each section plays a different conversational role. Readmy tags where language departs from what that role calls for, independent of the heading labels an author chose.

It does not decide whether content is true, good or novel. It does not fact check, appraise data or make accept and reject calls. It indexes structural markers so reviewers can spend their attention on judgement.

01

Introduction

Frames the problem and states the contribution

02

Methods

Describes how the work was conducted

03

Results

Reports what was observed

04

Discussion

Interprets, contextualises and argues

Marker tags / section roles
ClaimDiscussion
Data mentionResults
Citation flowIntroduction
Figure mentionResults
Text anomalyMethods
Raw text only. Figures and tables are never ingested.

Core attributes

Four primary markers, one wider mesh of checks.

The four primary marker types carry the map. Around them sits a wider mesh of structural checks that keep the reading consistent across a whole manuscript.

Claim Markers

Locates where assertion type language appears and in which section, surfacing argumentative density across the manuscript.

Data Mentions

Tracks where empirical and numeric statements sit, separating evidence bearing sentences from interpretive ones.

Citation Tracking

Maps where a reference first appears, where it recurs, and whether it resolves to the reference list.

Section Role Mapping

Maps text by its function in the argument, independent of the heading labels an author chose.

  • Claim to support linkage
  • Hedging and modality shifts
  • Terminology first appearance
  • Self citation density
  • Figure and Table sequence
  • Numbering integrity
  • Acronym first use
  • Text anomalies
  • Placeholder language
  • Generated language patterns
  • Section length ratios
  • Reference density
  • Cross module consistency
  • Near duplicate paragraphs

Ingestion is raw text only. Figures, images and tables are never ingested and never surfaced, so a figure exists to Readmy solely as the mention that calls it in the text.

The output

The ReadMy Index is a map, not a score.

Readmy never reduces a manuscript to one number. It returns a structural map and a QC checklist overlaid on the document, showing where markers concentrate.

Heavy concentration in the Discussion against a light Introduction is expected, so it is not itself a concern. The Index gives reviewers structure, not a verdict.

Clear Marker present
Readmy Index / tagged sentences

Introduction

Section role

Methods

Section role

Results

Section role

Discussion

Section role

Marked sentence Original manuscript Annotated copy

Privacy by design

Read once, blinded.
Never held as written.

Text is ingested a single time to tag structural features. Everything after that, the mapping, the Index and the calibration, runs on a decoy model rather than the author's language.

A decoy retains section and sentence structure while masking the words. The author's wording is never changed, never rewritten and never retained.

01

Ingested and tagged

Text is read once, blinded, to tag structural features. Nothing is rewritten and nothing is altered.

On-device processing

02

Abstracted to a decoy

Section and sentence structure is retained. The content is scrambled, and legible text is discarded.

03

Mapped and indexed

All downstream mapping and indexing runs on the decoy, never on the author's language.

Where Readmy differs

What Readmy is, and what it is not

Other tools invite anyone to review any paper and return a score without saying what was measured. Readmy takes the opposite position, in public.

What Readmy is

  • Maps argumentative structure section by section
  • Tags claim language, data mentions and citation flow
  • Returns the ReadMy Index as a structural map
  • Runs on decoy models rather than the author's words
  • Built for journal and editorial office workflows
  • States exactly what it measures and what it ignores

What Readmy is not

  • A fact checker or a credibility judge
  • A verdict on novelty, quality or significance
  • A pass or fail score, or an accept or reject call
  • A store of legible manuscript text
  • A consumer tool open to anyone with a PDF
  • A black box, since every marker traces to its source

How it works

From manuscript to structural map

1

Submit through the workflow

A manuscript enters via the editorial office. Structural features are ingested and tagged once: claims, data, citations and section roles.

2

Structural mapping

Tagged features are mapped against the role each section performs, producing concentration patterns and consistency signals.

3

The ReadMy Index

A structural map and QC checklist returns to the reviewer, overlaid on the document and traceable marker by marker.

Request access

Provisioned for journals and editorial offices by request.

Readmy is deployed under institutional agreement rather than individual sign up. If your journal or editorial office wants ethical AI peer review inside its workflow, we would like to hear from you.

Why we are building it

Institutional enquiry

Contact our editorial team
  • Onboarding for journal workflows
  • Editorial office integration support
  • Privacy and handling documentation