AIVO DIVM — Data Integrity & Verification Methodology (v1.0)

DIVM is the data trust backbone of the AIVO Standard™ — a governance-grade framework for AI Visibility Assurance.

It defines auditor-grade reproducibility and verification rules for visibility metrics derived from LLMs:

• Reproducibility thresholds: CI ≤ 0.05, CV ≤ 0.10, ICC ≥ 0.80
• Evidence metadata: model name/version, timestamp (UTC), locale, prompt fingerprint
• Replay Harness specification (independent re-run of measurements)
• DIVM Compliance SDK/API (evidence ingestion, verification, reporting)

Citable record (Zenodo DOI): 10.5281/zenodo.17428848

---

Repository Layout

docs/ DIVM_Methodology_v1.0.pdf CHANGELOG.md api/ openapi.yaml examples/ evidence_payload.json verification_report.json schema/ evidence-schema-v1.json report-schema-v1.json governance/ LICENSE.md CODE_OF_CONDUCT.md CONTRIBUTING.md

---

Getting Started

• See docs/DIVM_Methodology_v1.0.pdf for the full methodology.
• Use api/openapi.yaml to integrate the DIVM Verification API.
• Validate submissions against schema/*.json.

---

License

• Methodology content is released under CC BY 4.0.
• Code samples, schemas, and SDK are licensed under MIT, unless stated otherwise.