Microsoft Fabric IQ: ontologies for Data Agents and Operations Agents
Microsoft Fabric IQ is a new Fabric workload that maps OneLake tables, Power BI semantic models and real-time Eventhouse streams to shared business entities. Its Ontology item, currently in preview, lets Fabric Data Agents and Operations Agents work with business concepts instead of raw schemas. Read on if you are deciding whether to model business meaning before connecting LLMs to your data.
Key points
- Ontology (preview) declares entity types, properties, relationships, natural-language rules and metrics over Lakehouse, Warehouse or Eventhouse tables.
- Generating an ontology from an existing Power BI semantic model carries DAX measures over as ontology metrics.
- RDF or OWL files can be imported for industry domain definitions.
- The built-in Graph lets agents run multi-hop traversals instead of hand-written multi-join SQL.
- Fabric Data Agents can be published via MCP to Microsoft 365 Copilot, Microsoft Foundry or Copilot Studio.
- The Ontology Agent is proposal-first: review every schema change before it is committed.
Microsoft has introduced Fabric IQ, a business data and operational intelligence workload inside Microsoft Fabric, to tackle a common failure: LLM agents querying raw warehouse tables tend to guess joins and misread cryptic column names. It rests on three layers: OneLake for unified data through shortcuts and mirroring, Power BI semantic models for curated KPIs and DAX measures, and an Ontology item that declares entities, relationships, rules and metrics bound to live tables and Eventhouse streams.
A published ontology also yields a built-in graph, letting agents trace multi-hop chains, such as an order to a shipment, a temperature sensor and then a cold chain breach, without hand-written joins. Fabric Data Agents answer natural-language questions and can be published through MCP to Copilot surfaces, while Operations Agents watch Eventhouse telemetry and trigger governed workflows. The Ontology item is still in preview, so plan accordingly.
Real-world use
- A supply-chain team tracing which shipments are affected when a cold storage sensor drifts out of range.
- An analytics team already maintaining Power BI semantic models that wants agents to answer with the same KPI definitions.
- An operations team that needs alerts on SLA breaches with governed follow-up actions.
Why it matters
If your data already lives in OneLake and Power BI, the ontology becomes the single place to define business meaning for both dashboards and agents. Expect to design entity and relationship models before wiring LLMs, and keep in mind the feature is still in preview.