Fabric IQ and Ontology: Giving Data a Shared Business Meaning

Every organisation faces the challenge of defining what counts as an active customer, how revenue should be measured, and whether sales should include returns. This ambiguity results in wasted time and inconsistent reporting. With the introduction of AI agents, this issue becomes even more critical, as these agents can only work with the definitions available to them, and inconsistent definitions lead to inconsistent answers.

Databricks, Snowflake, and Microsoft have all taken steps to address this challenge. Databricks released Unity Catalog Business Semantics, Snowflake introduced Semantic Views, and Microsoft is offering Fabric IQ, a workload within Microsoft Fabric that focuses on Ontology and semantic models. These tools aim to provide a single trusted source of metric definitions for analysts, engineers, and AI agents.

This article explores what Fabric IQ and Ontology are, how they differ from existing semantic models, the business value they provide, and the trade-offs of the platform. The article builds on a previous piece about Microsoft Fabric, which covered the platform's approach to unifying data across engineering, analytics, and data science on a single platform.

Fabric IQ extends the unification of where data lives, bringing business context into the platform through Ontology and semantic models, both built on OneLake. Semantic models provide trusted metrics for reporting, while Ontology defines the shared business language behind those metrics. Ontology is a structured description of business concepts and their relationships, with each concept carrying defined properties and relationships to other concepts.

In Fabric IQ, an ontology is a representation of the concepts that make up a business and their relationships. It provides a level of abstraction above a semantic model, describing what a concept means independently of how the data is stored. Ontologies can be generated from existing Power BI semantic models, providing a starting point for organisations with mature reporting.

Ontology is built from entity types, entity instances, and properties and relationships, all linked through a queryable graph. This graph allows relationships to be traced rather than hidden in join logic. Ontology brings business context into AI agents, providing a structured map of the business to reason over, leading to more consistent and explainable responses.