Microsoft is steadily repositioning Fabric from a unified analytics platform into a governed context layer for enterprise AI. An agent answering a revenue question needs to know which measure applies, how customers and transactions relate, and what information it may use. At FabCon and SQLCon in Barcelona, Microsoft showed how Fabric IQ, OneLake, Power BI semantic models, ontologies, and governance controls could supply that shared business context across applications and agents.

Fabric IQ Makes Business Context Reusable

Fabric IQ is the clearest expression of Microsoft’s direction. The announcement brings together data from OneLake, trusted metrics from Power BI semantic models, and operational context from ontologies and real-time intelligence. Microsoft also introduced an agentic experience for creating ontologies from semantic models, OneLake data, and context documents, with the ability to carry established Power BI measures into those ontologies. Modelers can also define condition-action rules in the ontology (in preview) through Fabric Activator so that alerts, workflows, and updates fire when business conditions are met. Users can connect an ontology to Microsoft’s operations agent, which monitors real-time data and recommends business actions. The result is a shared layer of business meaning and governed actions that can ground Microsoft Copilot, Foundry, GitHub Copilot, Fabric agents, and custom applications.

OneLake Extends From Data Sharing To Context Sharing

Microsoft is expanding OneLake through Google BigQuery mirroring, bidirectional integration with Salesforce Data 360, and deeper integration with Dynamics 365 Business Central. The more significant announcement is IQ sharing, now in preview, which can share governed tables and files alongside agent instructions and RDF ontologies. Microsoft is therefore extending OneLake from connecting distributed data toward distributing reusable business context across teams, partners, and AI systems.

Power BI And Fabric Apps Turn Context Into Operational Software

The new agentic app-creation experience in Power BI Desktop starts with a trusted semantic model and lets users describe, generate, edit, and publish purpose-built applications using natural language. These applications can accept inputs, write back data, retain shared state, and support operational workflows. Fabric Apps, in preview, provisions the back end for these applications, including generated GraphQL APIs, hosting, and Microsoft Entra ID authentication. The announcement shows how Microsoft intends to turn governed business context into applications, rather than leaving it inside reports.

Fabric Agents Connect Context With Execution

Microsoft deepened the connection between database agents, Fabric data agents, and Fabric IQ ontologies so that answers can be grounded in defined business concepts and relationships. Database agents continuously observe and monitor the health of the operational databases across the estate, and the context from all the databases is exposed in the Database Hub in Fabric. The agent-building experience can use database context, semantic models, glossaries, documentation, and data dictionaries, while operations agents add root-cause analysis and preapproved actions with human oversight. The agents go further by planning and executing long-running work such as migrations, lakehouse modernization, optimization, and data preparation within permissions and guardrails set by engineers. These announcements position Fabric as a platform where context informs both reasoning and controlled execution.

Governance And Operations Complete The Context Layer

A context layer must remain observable and controllable as more applications and agents depend on it. Microsoft announced new Fabric observability capabilities in preview, including a unified view of workspace telemetry, expanded monitoring of jobs and capacity, Fabric Activator-powered alerts, and AI-assisted investigation through operations agents, which have been generally available since June 2026. Database Hub and database agents extend that management approach across the database estate. Together, these capabilities support Microsoft’s larger proposition: Fabric is becoming the governed environment where enterprise data, business meaning, operational signals, applications, and agents can work from the same context.

What Enterprise Leaders Must Do

Microsoft’s announcements make business context an architectural and management responsibility. Leaders need to decide where shared context will improve AI outcomes, who will define it, and how it will remain governed across platforms. Start with these actions:

  • Build knowledge and context-layer literacy first. Establish a common understanding of semantic models, ontologies, knowledge graphs, metadata, retrieval, and policy controls. You must understand the role each plays in helping AI interpret the business accurately.
  • Choose decisions and workflows that need better context. Focus on a business domain with identifiable users, trusted data, established metrics, and measurable outcomes. Use the first implementation to test whether shared definitions improve answers, decisions, or execution.
  • Put domain experts in charge of business meaning. Context cannot be defined by data and technology teams alone. Bring domain experts together with data architects, knowledge specialists, governance leaders, security teams, application developers, and AI practitioners to jointly define entities, relationships, metrics, policies, permissions, and actions. Build a shared understanding of semantic models, ontologies, metadata, and retrieval as the team works.
  • Define the logical architecture before selecting technology solutions. Specify how data, semantic models, ontologies, documents, real-time signals, governance controls, applications, and agents should interact. This prevents the architecture from being determined by the capabilities of a single platform or product.
  • Connect context across platforms before silos take hold. As hyperscalers and business applications build separate context graphs, enterprises need a shared layer that gives people and AI agents consistent definitions of entities, relationships, metrics, and rules. Microsoft’s support for RDF ontologies and Apache Ossie underscores the value of standards-based interoperability. Assess whether definitions, relationships, and metadata can move across analytics, AI, and business applications without becoming dependent on one platform.

Watch for our upcoming research on this topic. Forrester clients can schedule an inquiry or guidance session with me or Boris Evelson to understand why the context layer is becoming a differentiator in the AI stack and what it means for their architecture and vendor choices.

Share