Sanctuary of Ma'at Research Institute
The paper is weighed against the feather.
No exceptions.
Multi-Agent Architecture
Guardian routes every manuscript through four independent specialist agents simultaneously. Each agent is purpose-built for a distinct fraud vector. The Orchestrator synthesises all findings into a single evidence-backed verdict: CRITICAL, FLAGGED, or COMPLIANT.
Competitive Landscape
The research integrity tooling market is fragmented. Most solutions address one dimension. Guardian occupies the multi-agent orchestration layer that currently has no direct equivalent — including paper mill detection that our nearest competitor publicly admits is still in development.
| Tool | Their Gap | Guardian's Edge |
|---|---|---|
| iThenticate / Turnitin Plagiarism |
No statistical, ethics or citation fraud detection | Detects fabricated data — what iThenticate cannot see |
| StatCheck Statistical QA |
Single-dimension, no orchestration | Three additional agents beyond statistical checks |
| Proofig / ImageTwin Image Analysis |
Visual only — no textual fraud patterns | Complementary — Guardian covers the text layer |
|
Morressier
Paper Mill: In Development Research Integrity Platform |
Publicly admitted paper mill detection not yet built (July 2023) | Guardian ships paper mill detection today |
| GPT-based reviewers AI Writing QA |
No fraud-specific training, hallucination risk | Rule-based + pattern agents — fully auditable, zero hallucination |
Live Demo
Paste any manuscript excerpt below and Guardian will run it through all four specialist agents. Every finding includes the specific evidence and the historical retraction case it matches.
Guardian identifies risk signals based on patterns from 300+ documented retractions. All verdicts are probabilistic findings requiring human review — not legal determinations. Built on Complete.dev · Complete AI Hackathon 2026