Questions the strategy answers
6 of 6 questions
Not because the old way was wrong — because the consumer changed. Dashboards needed documentation; AI agents need deterministic, machine-readable contracts. Governance has to produce trust that a machine can act on.
Data evaluated against ten characteristics: stewardship, metadata enrichment, quality, lineage, access controls, compliance, bias awareness, change transparency, fitness for purpose, and continuous certification. AI does not fix poor data — it amplifies it.
It is federated and non-invasive. Central policies and tools are executed by the domains closest to the data, aligning with existing work behaviors instead of creating new bureaucracy — so governance accelerates the business rather than blocking it.
Metadata is descriptive — the what, where, when, and who, living in catalogs and glossaries. Context is interpretive — the why and how — and it lives in people. Without governing context, AI can be confidently wrong.
It is an open question the POC is testing. A modern catalog is structurally a knowledge graph and may become the semantic layer once exposed to agents via MCP. We are evaluating dbt, AtScale, and Cube on a weighted scorecard.
A cross-functional body of Business Owners, Data Stewards, Domain Owners, Data Product Owners, and Technical Custodians, facilitated by the EDG team. It meets bi-weekly, seeks consensus, and escalates unresolved issues to the quarterly Executive Governance Council.
