Evidence-backed text origin review workspace

Reduce disputed text-origin decisions while preserving a defensible evidence record. For educators, admissions reviewers and editors checking whether written text was AI-generated or human-written, convert submitted documents, author statements, permitted source material and review policy into reviewer-signed text-origin evidence reports linked to the reviewed submission. The benefit is a testable hypothesis, measured through reviewer-agreed origin decisions per review hour and upheld decisions after appeal; do not assume that AI output alone produces business value.

Confirm the buyer's problem and scope, collect submitted documents, author statements, permitted source material and review policy, then follow this sequence: 1. Detect whether supplied text is AI-generated or human-written. 2. Produce a calibrated likelihood score with stated uncertainty. 3. Highlight suspected sentences for detailed review. 4. Generate a structured report with signal breakdowns and insights. Resolve uncertain cases with qualified reviewers, sign reviewer-signed text-origin evidence reports linked to the reviewed submission, and measure reviewer-agreed origin decisions per review hour and upheld decisions after appeal against a documented baseline.

How the AI works

Use AI to interpret permitted inputs, suggest structured mappings and generate candidate outputs for the three stated task modules. Use deterministic code for arithmetic, schema validation, hard constraints and reproducible tests. Review source-linked explanations and uncertainty before accepting results. Detection signals are probabilistic; final origin judgments and academic consequences remain human. A model suggestion is never a verified fact, professional decision or authorization to act.

Safeguards

Preserve author voice, source attribution, quotation accuracy and usage permissions. Reviewers approve substantive origin judgments and academic consequences. One review policy and one document type; final origin judgments and academic consequences remain human. Keep all consequential actions under authorized human control and do not fabricate missing inputs, permissions, professional judgments or market evidence.

What to build first

Pilot scope: One review policy and one document type; final origin judgments and academic consequences remain human. Implement one approved input format, a bounded representative case set and the first two task modules: detect whether supplied text is AI-generated or human-written; produce a calibrated likelihood score with stated uncertainty. Support the third module with operator review: highlight suspected sentences for detailed review. Include source references, corrections, basic organization access, sign-off states, export and value measurement. Use managed operator assistance for unresolved exceptions. The cost estimate covers this narrow prototype, not unrestricted multi-tenant scale, complex production integrations, specialist certification or physical operations.

What it can connect to

Institution-owned submission systems, authorized author statements and permitted research sources. Cloud document storage, learning-management-system import/export and reporting destinations. Start with file exchange and validate destination specifications before promising direct publishing. Start with authorized file exchange. Validate current provider access, usage rights and schema behavior before promising a connector.

The screens in detail

Primary screens: Submission intake and policy, Editable review workspace, Signed report and delivery. Use a thumbnail gallery for submissions, a large central text canvas with sentence-level highlighting, and a right-hand panel for signals, sources and comments. Let users compare detector runs side by side. Display draft, changes requested and signed states. Provide a shareable report link with comments anchored to the relevant passage. Make the task-specific outcome reviewer-signed text-origin evidence reports linked to the reviewed submission visible beside its evidence, review state and value baseline.