Microsoft Fabric IQ: Enterprise Ontology as the Brain for AI Agents
🔄 Update — 15 June 2026: Microsoft Fabric partners with Replit for AI-Powered App Building
Microsoft and Replit have announced a partnership to allow developers to build and deploy enterprise data applications directly within Microsoft Fabric. Using Replit’s AI-assisted development environment and the new open-source Rayfin SDK, developers can quickly create vibe-coded apps connected to OneLake.
What’s new?
- Replit Integration & Rayfin SDK: Developers can write and deploy applications directly inside their Fabric tenant using the Rayfin SDK and CLI, leveraging Replit’s AI agents.
- OneLake Integration: Applications access Fabric data directly in OneLake, removing the need for external database hosting and complex data pipelines.
- Enterprise Governance: Built-in integration with Microsoft Entra ID and Microsoft Purview ensures all developed apps remain secure and compliant within the enterprise perimeter.
Why this adds to the article
This integration bridges the gap between structured enterprise data in Fabric IQ and custom AI applications. By simplifying the application development workflow, organizations can now rapidly build agentic tools that leverage their central Enterprise Ontology.
🔄 Update — 10 June 2026: Microsoft Fabric Integrates Native EPM/CPM Planning
Microsoft has introduced Planning in Fabric (part of Fabric IQ) as an integrated EPM/CPM capability. This allows organizations to manage budgets, forecasts, and scenario models directly on the same OneLake data foundation, eliminating the need to transfer data to third-party tools. Additionally, new administrative features in the Workload Admin Portal and updated SKU allocations simplify capacity management.
What’s new?
- Native EPM/CPM Planning: The integration of Plan, PowerTable, and Intelligence within Fabric IQ enables seamless budgeting and scenario modeling.
- No Data Movement: Because planning is built directly on OneLake, latency and reconciliation with external tools are eliminated.
- Admin Enhancements: The new Workload Admin Portal provides advanced monitoring capabilities for capacity and item-level configurations.
- Preview SKU Model: During the preview phase, planning features can be explored using existing capacities (F-SKUs) without additional billing.
Why this adds to the article
Native EPM/CPM planning extends Fabric IQ’s Enterprise Ontology with forward-looking financial and scenario data. This allows AI agents to not only analyze historical data but also directly participate in forecasting and budgeting workflows.
Summary
Microsoft has introduced Fabric IQ, a new intelligence layer for the Fabric ecosystem. At its core is the “Enterprise Ontology,” which acts as a semantic bridge between raw data and AI agents. This allows organizations to interpret data from various sources consistently and elevate AI-driven analytics to a new level of reliability.
What happened?
As part of the evolution of Microsoft Fabric, Fabric IQ was introduced (currently in preview). It is a layer that sits on top of OneLake, not just storing data but assigning it business meaning. Through integration with Azure AI Foundry, AI agents can now directly access this ontology to answer complex business questions in natural language.
Why it matters
The biggest hurdle for AI in the enterprise is often a lack of context. Without understanding the relationships between customers, orders, and products, LLMs (Large Language Models) are prone to hallucinations. Fabric IQ solves this problem by defining a “language of the business” that is understandable to both humans and machines. This transforms AI agents into true business partners capable of reasoning autonomously across data sets.
Evidence
- Official Documentation: Microsoft has released detailed guides on configuring Fabric IQ and integrating it into agent workflows.
- Integration: The link with the Model Context Protocol (MCP) in Azure AI Foundry demonstrates the strategic focus on agentic workflows.
- Features: Features like NL2Ontology (Natural Language to Ontology) enable direct querying of complex data models without SQL knowledge.
Analysis
Fabric IQ marks Microsoft Fabric’s transition from a pure data platform to an intelligence platform. While classic semantic models are primarily optimized for reporting, the Enterprise Ontology is designed for “reasoning.” This is a crucial step for scaling AI applications, as it eliminates the tedious manual data preparation for every individual bot.
Practical Takeaways
- Central Definition: Organizations should start defining their core business terms centrally in the ontology rather than maintaining silos in Power BI or SQL databases.
- Agent Readiness: Fabric IQ is the prerequisite for safely deploying AI agents on enterprise data without them losing their way.
- Governance: Since Fabric IQ inherits Fabric’s security and compliance rules, data sovereignty is maintained.
Open Questions
- Pricing Model: The exact costs after the preview phase are not yet fully known.
- Performance: How does the ontology query scale with extremely large and complex global data models?