Index Methodology

GRandCIndex monitors a curated set of high-authority public sources — including regulatory bodies, governmental institutions, standards organisations, intergovernmental forums, and recognised research institutions. Inputs are collected through a combination of automated retrieval and editorial sourcing. Raw inputs enter the platform as candidate items and are not visible publicly.

Candidate items are classified against the GRandCIndex taxonomy. Classification assigns thematic tags, jurisdictional scope, signal strength on a four-tier ladder (Weak / Moderate / Strong / Structural), directionality (Escalating / Steady / De-escalating / Reversing), and a time horizon. Classification structures the item; it does not publish it.

Every classified item passes through editorial review before it can become a public signal. Reviewers assess source authority, relevance to AI governance and accountability, novelty against existing intelligence, and escalation potential. Items that do not meet the threshold for decision relevance are rejected or held.

Approved items move through a controlled pipeline: In Review → Approved → Published. A signal may be published individually, incorporated into an intelligence brief, both, or held privately for analytical context. Brief generation is a separate editorial act, not an automated consequence of signal approval. There is no auto-publish path from ingestion to homepage.

Where readiness or exposure indicators are presented, they are analytical estimates derived from observable signal patterns — including convergence across themes, jurisdictional pressure clustering, signal strength distribution, and directionality trends. They are intended to support executive orientation, not to measure organisational compliance.

The underlying classification heuristics, scoring weights, model configurations, prompt engineering, taxonomy structures, and analytical algorithms used by GRandCIndex are proprietary to GRandC Consulting LLC. They are intentionally not disclosed. Published outputs are designed to be interpretable on their own terms without requiring exposure of internal mechanics.