BlueHomer
The vocabulary of governed intelligence
Short, unambiguous definitions of the terms that come up when an organization tries to use AI on data it has to be able to defend.
All terms
Governed AI
Governed AI means every AI action is authorized before it runs, executed under a real identity, approved when privileged, and recorded in verifiable evidence.
Read the definition →LLM gateway
An LLM gateway is a mandatory control point between enterprise systems and AI models that applies policy, protects data on the way out and scans responses on the way back.
Read the definition →Attribute-level lineage
Attribute-level lineage traces an individual attribute — with the transformation rules applied to it — from a dashboard number back to its physical source column.
Read the definition →Data lineage
Data lineage is the recorded path data takes from origin to use. It exists at table, column and attribute level, and only the finest level explains a number.
Read the definition →Data catalog
A data catalog is an inventory of an organization’s data assets enriched with metadata, business meaning, ownership and certification so people and agents know what to use.
Read the definition →Semantic layer
A semantic layer is the business-friendly definition of what data means, authored once and referenced everywhere, so every tool and every AI answer uses the same meaning.
Read the definition →Text-to-SQL
Text-to-SQL converts a plain-language question into a database query. It fails when the model must guess business meaning it was never given — producing confident, wrong results.
Read the definition →Augmented analytics
Augmented analytics applies AI to automate data preparation, insight discovery and explanation — shifting analysts from building queries to validating and interpreting.
Read the definition →Conversational BI
Conversational BI lets people ask data questions in natural language and get answers back in context, without writing queries or waiting for a dashboard to be built.
Read the definition →Agentic ETL
Agentic ETL is data integration where an AI agent proposes the mapping, transformation and schedule, and the platform enforces permission, approval and audit.
Read the definition →Tamper-evident audit log
A tamper-evident audit log is append-only and hash-chained, so any alteration or deletion breaks the chain and can be detected independently of the system that wrote it.
Read the definition →ABAC (attribute-based access control)
RBAC grants access by role; ABAC decides using attributes of the user, data and context. Together with row and field controls, they reach the data rather than just the API.
Read the definition →