Skip to main content
fabric iq vs purviewfabric iq ontologyfabric ontology vs purviewdoes fabric iq replace purview glossarypurview business glossarypurview data governance retirement

Fabric IQ vs Purview: Ontology or Business Glossary?

A Fabric IQ ontology is a machine-usable model of business entities bound to Fabric data for AI agents, still in preview in November 2026, while a Purview business glossary holds definitions that people own and approve. They solve different problems, and Microsoft's Purview transition guidance does not map glossary terms to Fabric IQ at all.

The question comes up because the timing overlaps. Microsoft announced on November 12, 2026 that Purview Data Governance, including the Unified Catalog and its glossary, reaches end of support on November 12, 2027. At the same time, Fabric IQ ontologies are the part of Fabric that talks most about shared business vocabulary. This page sets out what Fabric IQ is, according to Microsoft's own documentation, how an ontology differs from a business glossary, and why most teams need both.

TL;DR

Fabric IQ was introduced in preview at Microsoft Ignite in November 2025 and declared generally available at Build in June 2026, although Microsoft Learn still labels the IQ workload and its ontology item as preview. Its ontology item is still in preview, with general availability expected "in the coming months". An ontology defines entity types, properties, relationships, rules and metrics and binds them to Fabric data so agents share one vocabulary. A Purview glossary holds terms with definitions, owners, domains and approval. Microsoft's transition guidance maps glossary terms to tags, descriptions and endorsement, and does not mention Fabric IQ. There is no documented import from a Purview glossary into an ontology. Treat them as complementary: the business agrees the definition in the glossary, the ontology models it for Fabric data.

What is Fabric IQ?

Fabric IQ is a workload in Microsoft Fabric that gives people and AI agents context about business entities and the data behind them. Microsoft introduced it at Ignite 2025, held November 18 to 21, 2025, where the Book of News described it as "now in preview" and as an extension of the semantic layer already provided by Power BI.

Microsoft Learn's Fabric IQ overview places it inside Microsoft IQ, next to Work IQ, Foundry IQ and Web IQ. Fabric IQ has two core items, the ontology (preview) and the Power BI semantic model, built on data in OneLake. The IQ workload also groups planning, graph, data agents and operations agents.

The status of each part differs, which matters when you plan around it:

  • Fabric IQ as a whole was announced as generally available at Build 2026 on June 2, 2026, together with graph in Fabric and operations agents.
  • Ontologies were "expected to be generally available in the coming months" at Build. The ontology documentation, last updated in October 2026, still marks the item as preview.
  • Agent access to ontologies is in preview too: as context for Fabric data agents, per Microsoft's What's new list, and as knowledge sources in Microsoft Foundry, per the Build announcement. At Ignite 2026, Microsoft added IQ sharing for ontologies, also in preview.

What is a Fabric IQ ontology?

Microsoft defines the ontology item as "a shared, machine-understandable representation of your business". It is a modeling layer: it stores definitions and bindings, while the data stays in the systems that own it. The general idea is covered in What is an ontology in AI?; the Fabric version is built from these parts:

  • Entity types such as Customer, Shipment or Machine, each with a name, description, properties and identifiers.
  • Relationships between entity types, such as "Customer places Order".
  • Metrics carried over from DAX measures in a Power BI semantic model, which stays the owner of the formula.
  • Rules written in natural language and linked to entities, so agents can use the same business logic.
  • Data bindings to supported Fabric sources: semantic models, lakehouses, eventhouses, warehouses, SQL databases, mirrored databases and supported OneLake constructs such as shortcuts.

An ontology can be built from scratch, generated from existing semantic models, drafted with the ontology agent, or imported from RDF, Turtle or OWL files. Metadata on entity types can include descriptions, synonyms and custom key-value attributes. Graph in Fabric is an optional layer for questions where the path between entities matters. For how this relates to a semantic layer, see Ontology vs semantic layer.

If your team experimented with Fabric IQ early, note that the first ontology experience retires on January 31, 2027, and items built in it need to be copied to the new experience.

Fabric IQ ontology vs Purview business glossary

Both deal with business meaning, so they are easy to confuse. They differ in who they serve, what a single entry is, and how changes are controlled. The Purview column covers both the classic glossary and Unified Catalog terms, as documented on Microsoft Learn; the ontology column reflects the preview documentation as of November 2026.

AspectPurview glossaryFabric IQ ontology
Main purposeAgree and publish what business terms meanGive agents and apps a model of the business bound to data
Basic unitA term with a definitionAn entity type with properties and relationships
Main audienceBusiness users, data stewards, analystsAI agents, dashboards, applications, model authors
OwnershipOwners and experts on each termFabric workspace and item permissions; no owner role on an entity type is documented
Change controlDraft and published states; approval workflows in the classic glossaryManually created named versions; ontology agent changes are proposed for review
Synonyms and hierarchySynonyms, acronyms, parent and related termsSynonyms in metadata; inheritance between entity types
ScopeGovernance domains, linked to assets in the Purview Data MapSupported Fabric data sources through data bindings
Business logicDescribed in the definition textRules in natural language, metrics from DAX measures
StatusRetiring; end of support November 12, 2027Preview; general availability expected

In short, a glossary term answers "what do we mean by active customer, and who decides?". An ontology entity type answers "which table holds customers, how do they connect to orders, and which measure counts the active ones?". The first is a governance record; the second is an engineering model that depends on the first being settled.

Business Glossary, Fabric IQ Ontology and Agents BUSINESS GLOSSARY, FABRIC IQ ONTOLOGY AND AGENTS BUSINESS GLOSSARY Definitions people agree Owners and approval Synonyms and domains What a term means FABRIC IQ ONTOLOGY Entity types, relationships Rules and metrics Bound to Fabric data How the data models it AGENTS AND APPS Fabric data agents Operations agents Microsoft Foundry agents Who uses it Agree the definition once in the glossary, then model it for Fabric data in the ontology. Ontology and agent access to it are in preview as of November 2026.
Click to enlarge

Does Fabric IQ replace the Purview glossary?

Not according to Microsoft's own guidance. The transition guidance published on November 13, 2026 maps "glossary, critical data elements, custom context" to tags, descriptions and endorsement in Fabric. Its advice is to "export and preserve terms and ownership", and it warns: "Do not claim full glossary or CDE parity." The guidance does not mention Fabric IQ or ontologies anywhere.

There is also no documented path from a Purview glossary into an ontology. The ontology import feature accepts TTL, RDF or OWL files, and only into an empty ontology item. During import, a label becomes the display name, a comment becomes the description and an alternate label becomes a synonym. In principle, you could convert a Purview glossary export into Turtle and import it. That is our inference, not a Microsoft procedure, and it runs into three problems:

  • Most glossary terms are not entities. "Customer" can become an entity type. "Net revenue" is a metric, "churn" is a rule or a calculation, and "GDPR personal data" is a classification. Loading every term as an entity type produces a model that agents cannot use well.
  • Governance fields have nowhere to go. Owners, experts, approval status and governance domains have no documented equivalent on an entity type. Custom key-value metadata could store them as text, but without ownership or approval behind it.
  • Scope narrows to Fabric. Bindings reach supported Fabric sources. Terms that describe data in SAP, Snowflake or Databricks lose their link to that data.

Microsoft's import documentation itself warns that a successful import "doesn't guarantee that all source semantics were retained". Keep the original glossary as the record of what was agreed.

See it in action

Dawiso as a Microsoft Purview alternative

Keep your Purview glossary terms with owners and approval workflows, and give AI agents access to them through MCP.

How a business glossary and an ontology work together

The glossary and the ontology sit at two steps of the same process. People first agree what a concept means and who owns it. Engineers and model authors then express that concept as entity types, relationships and bindings that agents can query. When the order is reversed and the ontology is built first, the definitions in it are whatever the model author assumed, and nobody in the business has signed them off.

  1. Agree the definition in the glossary. The term, its definition, synonyms, owner and approval live where business users can read and challenge them.
  2. Decide what the term is. An entity, a property, a metric, a rule or none of these. Only some glossary terms belong in an ontology.
  3. Model it in the ontology. Copy the approved definition into the entity type description, add synonyms, and bind it to the Fabric tables that hold the data.
  4. Keep a link back. Record the glossary term identifier in the entity type metadata, so a changed definition can be traced to the entities that use it.
  5. Review changes in both directions. A new entity type without a glossary term is a sign that a definition was never agreed.

The same pattern appears on other platforms. Databricks offers its own ontology for Genie, described in What is Databricks Genie ontology?, and each platform's ontology covers that platform's data. A glossary kept outside any single platform can feed several of them with the same definitions.

Where to keep the glossary after Purview

With Purview Data Governance retiring and Fabric offering tags and descriptions in place of a glossary, the agreed definitions need a home that keeps owners and approval. That is the role of a data catalog with a business glossary. Exporting the terms themselves is covered step by step in Purview glossary migration, and the wider Fabric mapping in Purview to Fabric.

Dawiso keeps business terms with owners, synonyms, domains and approval workflows in its business glossary, and the Dawiso customer success team runs the Purview glossary migration with you.

For agents, the Dawiso MCP server, available for Dawiso cloud customers, lets AI tools search catalog objects, read their details and trace lineage, with the tools available depending on the user's role. Dawiso catalogs Microsoft Fabric and Power BI together with Databricks, Snowflake and other sources, so the same approved definitions can sit next to the data they describe and be copied into a Fabric IQ ontology where you use one. The background on the retirement is in Microsoft Purview Data Governance is retiring.

Sources

See it in action

Keep your glossary when Purview retires

Dawiso keeps business terms with owners, synonyms and approval workflows, and serves them to AI agents through MCP.

A cookie a day keeps bad UX away.

We use cookies to personalize content, ads and to analyze our traffic. We also share information about your use of our site with our advertising and analytics partners who may combine it with other information that you've provided to them or that they've collected from your use of their services. By clicking "Accept All", you allow us to use cookies for analytics and ads via Google Tag Manager. You can also customize cookies.

Customize Consent Preferences

We use cookies to personalize content, ads and to analyze our traffic. We also share information about your use of our site with our advertising and analytics partners. Privacy Policy

Necessary cookies allow core website functionality such as user login and account management. The website cannot be used properly without strictly necessary cookies.

Functionality cookies are used to remember visitor information on the website, eg. language, timezone, enhanced content.

Analytics cookies are used to see how visitors use the website, eg. analytics cookies. Those cookies cannot be used to directly identify a certain visitor.

We use Microsoft Clarity to see how you use our website (including heatmaps and session replays) so we can improve it. By using our site, you agree that we and Microsoft can collect and use this data. See our Privacy Policy for details.

Advertisement cookies are used to identify visitors between different websites, eg. content partners, banner networks. Those cookies may be used by companies to build a profile of visitor interests or show relevant ads on other websites.