Governance, Security & Compliance

Governance is not a layer you bolt on after a query runs - it is structural to how the query gets compiled in the first place. In a semantic execution layer, RBAC, ABAC, and row/column-level predicates are evaluated at compile time, before any SQL touches the warehouse. Unauthorised queries fail compilation. Filtered-out rows are never read.

This collection covers the governance-by-design model that production AI demands. Posts explore fine-grained data access control as a compile-time concern rather than a runtime patch; how data authorisation breaks when meaning, identity, and policy are stored in three different systems; and what HIPAA-grade conversational analytics looks like when every query is provably grounded in the semantic graph and policy-aware before it executes.

You'll also find pieces on multi-tenant semantic isolation, audit trails that are point-in-time reproducible, and the regulatory consequences of letting an LLM emit ungoverned SQL against a regulated dataset. The argument running through them: governance should be a property of the compiler, not a wrapper around the answer. When policy is structural, compliance becomes the default, audit becomes free, and every AI agent inherits the enterprise's existing controls without a single new line of access logic.