If you haven’t already, check out Arun Ulag’s hero blog “Microsoft Build 2026: Building Agentic Apps with Microsoft Fabric and Microsoft Databases” for a complete look at all of our Microsoft Build announcements across our Fabric and database offerings.
_______________________________________
Every leadership team is asking the same question: how do we win in the age of AI?
The answer is no longer just better models. It is whether your people and AI agents can operate together, in real time, sharing an always-current understanding of the state of the business. Today, that rarely happens. Signals are everywhere, but they are fragmented across systems, interpreted differently by teams, and acted on without full context.
Closing that gap requires an operational foundation that unifies data, signals, and context into one system. Microsoft Fabric provides that foundation. It brings together OneLake as the unified data layer, Real-Time Intelligence as the event and action plane, and Fabric IQ as the shared context layer to enable real-time apps and agents to operate from a consistent understanding of the business.
At Build, we are announcing the general availability of Fabric IQ, making this shared context layer available in production. This includes the general availability of Operations Agents and Graph, with Planning reaching general availability later in June. Ontology, the operational context layer that grounds agents in business meaning, continues in preview with expanded capabilities, improvements, and integrations across Fabric, Foundry, and Microsoft 365 Copilot.
Figure: The operational foundation for the modern enterprise: Fabric IQ, Real-Time Intelligence, and OneLake.
Why this matters now
According to a June 2026 Forrester report, Combine Semantics, Ontology, and Knowledge Graphs For AI-Ready Data, “AI fails without context-rich data. No matter how well modeled, data without the context provided by shared semantics and mature ontologies will fall short for agentic AI use cases. Enterprises that embed semantics across their tech stack unlock exponential returns: faster time to insight, reduced waste, and boosted ROI.”
Three shifts are making operational intelligence both urgent and achievable:
- Signals are everywhere. Enterprises are rapidly becoming instrumented. Applications, devices, and operations now continuously emit telemetry. The problem is no longer access to signals; it’s making sense of them in a consistent, business-aligned way.
- Time to act is shrinking. In modern operations, delays are costly. Detecting an issue is not enough if the response comes too late. The expectation has shifted from understanding what happened to acting on what is happening now, while it still matters.
- AI can act but needs context. Agents can reason and execute, but their effectiveness depends on shared, reliable context. Without it, the same signal leads to different interpretations and inconsistent outcomes across teams and systems.
These shifts expose a gap in most enterprises. Real-time infrastructure can surface events, but it cannot keep decisions aligned across the business. Closing that gap requires a shared context layer, so every signal is interpreted the same way, and every action reflects the full operational picture.
Fabric IQ is the business context layer
Fabric IQ delivers that shared context layer. It connects data, business meaning, and live operations so decisions can be made with more context, in the moment. It is built on three connected layers:
- Unified data in OneLake. All enterprise data, including structured, unstructured, event, and graph, comes together in a single open foundation.
- Business Intelligence through semantic models. Data is defined once with consistent business meaning, so every team operates from the same logic.
- Operational Intelligence through ontologies and real-time signals. Live signals are interpreted in business context so decisions are made and executed in the moment.
Ontologies model the entities, relationships, and processes — devices, sites, incidents, contracts, customers, operations — that semantic models were never designed to represent, and that the agent action loop requires.
Figure: The three connected layers of Fabric IQ: unified data in OneLake, business intelligence through semantic models, and operational intelligence with real-time signals and ontologies.
But running a business requires more than what lives in systems alone. Decisions are shaped by how people work, collaborate, and apply both institutional knowledge and real-time web context. To fully operationalize intelligence, that same context must extend across the entire organization.
This is where Microsoft IQ comes in. It is the enterprise intelligence layer of the Microsoft stack delivering a shared, continuously updated understanding of your organization so every person and every agent operate from the same context.
Figure: The enterprise intelligence layers that make up Microsoft IQ: Web IQ, Work IQ, Fabric IQ, and Foundry IQ.
The enterprise intelligence layer spans:
- Web IQ brings real-time global context from the web.
- Work IQ captures how work actually happens across teams, roles, and workflows.
- Fabric IQ represents the state of the business across data and operations.
- Foundry IQ reflects reusable knowledge bases and authoritative documents.
Fabric IQ is where that enterprise intelligence layer is operationalized.
The operating loop for modern enterprises
In practice, this takes the form of a continuous operational loop: observe what is happening now, understand it in business context, decide the right response, and act with governance and precision.
Figure: The operating loop in motion: live signals flow in, teams and AI agents observe, analyze, decide, and act with business context, actions flow back out.
Fabric IQ enables this loop to operate consistently across the business, supported by a set of capabilities: Semantic Models, Ontologies, Digital Twin Builder, Graph, Planning, Data Agents, and Operations Agents. To go deeper on these capabilities, learn more in the Trusted AI starts with Microsoft Fabric: Unified real-time intelligence and IQ context blog post.
The loop only works if it is continuously powered by live signals and can trigger action in the moment. Real-Time Intelligence is the event and action plane that delivers those signals, ensuring decisions result in timely, coordinated action. To go deeper on the capabilities within Real-Time Intelligence, learn more in the Microsoft Fabric Real-Time Intelligence: A Leader in the 2025 Forrester Streaming Data Wave blog post.
Fabric IQ and Real-Time Intelligence operate as a single system: Real-Time Intelligence delivers the live signals while Fabric IQ provides the context and rules, enabling agents and applications to act in real time with shared understanding.
This is the shift from reacting to events to operating on them. But it depends on more than just real-time signals. Agents must retain context over time—learning from past events, tracking what changed, and building a consistent understanding of the system. That requires a fundamentally different approach to memory. We explore this in more depth in a separate article, Why your AI agent has amnesia and why forgetting is the fix, that covers how agents can retain and evolve context over time.
The operational loop in practice: Siemens Healthineers
The impact of Real-Time Intelligence within Fabric is most clear when applied to a real operational environment. At Siemens Healthineers, developers are building real-time, event-driven applications to monitor and support critical medical systems, where reliability and response time directly impact patient care, and clinicians depend on immediate insight – often while a patient is on the table.
To meet these demands, they are shifting from batch-based log analysis to continuous streaming with Fabric. Instead of diagnosing issues after the fact, signals from connected devices are processed as they occur, enabling faster issue detection, more precise responses, and reduced operational overhead.
As Dr. Werner Zirkel from Siemens Healthineers explains, real-time streaming is becoming foundational:
“Streaming is now becoming a necessary commodity. Systems need to operate on data in real time, and Fabric enables that shift. It ensures reactive services can pinpoint issues exactly where they occur, trigger actions like ordering spare parts, and help prevent downtime for customers, i.e. we can become more proactive which ultimately increases customer benefit.”
The outcome is clear:
- Faster response when conditions change.
- Consistent decisions across teams and systems.
- Reduced manual triage and intervention.
- Lower operational costs through automation and streamlined signal processing.
- Actions that are governed, auditable, and aligned to policy.
This is the loop in motion. Signals are observed, understood, and acted on in real time, changing outcomes in the moment.
What’s new in Fabric
At Build this week, we are announcing general availability of Fabric IQ, along with Operations Agents and Graph, with Planning reaching general availability later in June. We are also introducing targeted preview updates that make it significantly easier to put real-time, agent-driven scenarios into operation.
Fabric IQ updates
Operations Agents (Generally Available)
Operations Agents extend Fabric IQ into real-time execution. They continuously monitor live conditions, evaluate them against business rules, and recommend or execute actions. They run on the same governed substrate as the rest of Fabric — one security model, one governance framework, one data plane on OneLake — and they integrate natively with Real-Time Intelligence for streaming signals and with Microsoft Foundry for agent orchestration. Teams can put a single operational loop into production today and compound from there. This brings intelligence directly into the flow of operations, reducing manual intervention and enabling faster, more consistent decisions.
Graph (Generally Available)
Graph in Microsoft Fabric is the platform’s relationship-first data modeling engine, explicitly capturing how data is connected rather than leaving context to be inferred. Now generally available, it enables teams and AI agents to reason across relationships in real time and understand how changes ripple across the enterprise with full context. As part of Fabric IQ’s shared context layer, Graph works hand in hand with Fabric’s ontology to provide a consistent, governed view of complex relationships, forming the foundation for connected, AI-powered decision-making across domains.
Now, Graph delivers the scale, performance, and reliability required for production workloads, supporting massive, highly connected datasets while maintaining predictable query performance and operational stability. The engine is natively built on the Graph Query Language (GQL), providing rich, standards-based support for both deep traversals and broad pattern matching, so teams can express complex relationship logic with precision and confidence. Graph also aligns with Fabric’s enterprise promises by operating directly on data in OneLake and inheriting the platform’s security, governance, and compliance model without duplicating pipelines or weakening trust boundaries.
Planning (Generally Available later in June)
Planning brings insight into coordinated action. Teams can plan over the same trusted data, semantic models, and ontologies that power reporting, operations, and AI, creating faster, more precise plans with built-in context. Notably, plans are not static outputs. They can be written back into Fabric to drive execution, enabling closed-loop alignment. This is what extends Fabric into a unified decision platform, where planning and execution are not separate steps, but part of a continuous operational flow. To learn more about planning, check out the Introducing Planning in Microsoft Fabric IQ: From historical data to forecasting the future blog pos....
Ontology: Operational Context Layer for AI Agents (Preview)
Ontology capabilities, currently in preview, continue to expand, strengthening how business meaning is defined, shared, and applied across Fabric, Foundry, and Copilot Studio. These updates make it easier to model business entities, enforce consistent logic, and govern how data is interpreted and used across teams and agents. To learn more about ontology, check out the Fabric IQ: ontology blog post.
To quantify the business impact of Fabric IQ, we ran an internal benchmark comparing agents that use Ontology as a source with agents that don’t use Ontology as a source. Ontology-grounded agents produced 2.2× more "excellent" responses, +4.5-point improvement in context relevance, and achieved a 4.5× win rate in side-by-side comparisons and did so with 30% fewer tool calls. Without ontology grounding, agents can get lost, exploring the wrong paths and failing to reach the right answer. With ontology grounding, agents start with the map: clear entities and relationships that guide them directly to the correct result, more often and more reliably.
Figure: Based on a Microsoft-controlled A/B study using the RAG Triad framework results reflect 4,000 responses to 400 questions (Fabric IQ ontology-grounded vs. baseline). Scores are standardized and aggregated.
Fabric IQ integrations (Preview)
At Build, we are extending Fabric IQ full scope integrations, including ontology, into preview across Foundry, Copilot Studio, and GitHub Copilot and MCP surfaces. Fabric IQ enables operational context to be defined once and applied consistently across agent-driven experiences, Copilot environments, and developer tools, allowing agents and applications to reason and act with the same understanding, regardless of where they are built or run.
Figure: wide spectrum of agents available for no code, low code, and pro code developers.
Microsoft Foundry: For the pro-code developers, Fabric IQ ontology, data agents, and semantic models are available. Data agents enable access to raw data from Fabric. Semantic models provide curated meaning for analytics and reporting. Ontology (preview) adds the full operational context, defining entities, relationships, rules, and actions. Together, these layers enable developers to move from data access to actionable understanding across agent-driven applications.
Microsoft Copilot Studio: For low-code and no-code builders, Fabric IQ ontology and data agents are integrated into agent workflows, allowing agents to reason and act using consistent business meaning.
Microsoft 365 Copilot: Fabric IQ is expanding into Microsoft 365 Copilot through Cowork and BizChat (Frontier), bringing governed Fabric data directly into agent workflows, starting with Power BI. Agents don’t just surface insights. They track changes, generate updates, and drive next steps in real time, with execution grounded in governed and trusted data.
Agent and developer tools: Fabric IQ extends into agent-building ecosystems through MCP support, including Fabric IQ ontology (Preview), Power BI MCP (Preview), and Data agent MCP (Generally Available), enabling use across GitHub and other developer surfaces.
Real-Time Intelligence updates
Real-time actions and triggers – Dataflows and User-Defined Functions (Generally Available) and Copy Jobs (Preview)
Activator in Fabric continues to expand how actions are triggered from live conditions, including support for copy jobs, dataflows, and user-defined functions. These capabilities enable decisions made in real time to be executed reliably across operational systems. To learn more, check out What is Fabric Activator.
Dynamic thresholds and stateful conditions (Generally Available)
Activator now support more adaptive monitoring, allowing teams to define conditions that evolve over time and reflect real operational patterns. This improves precision and reduces noise, ensuring only meaningful events drive action.
Eventstream and Eventhouse enhancements (Preview)
Ongoing investments in Eventstream and Eventhouse improve observability, connectivity, and low-latency analytics. These updates make it easier to ingest, process, and evaluate signals at scale without custom infrastructure.
Expanded connector ecosystem (Preview)
New connectors continue to broaden access to enterprise systems, making it easier to bring operational data into Fabric and incorporate it into real-time workflows.
Real-time application experiences (Preview)
Enhancements across dashboards, embedding, and application integration make it easier to bring real-time insights and actions directly into apps and operational surfaces where decisions are made.
From insight to action
Winning with AI is no longer about experimentation. It is about execution. Leading organizations are already running parts of their operations in real time, connecting signals, context, and action and improving outcomes. The gap is widening.
Fabric makes this shift achievable. It brings real-time signals, shared context, and governed action into a single system, so teams do not need to stitch together tools or rebuild the stack for every use case. The advantage now comes down to time-to-value. The faster you operationalize intelligence, the faster you improve outcomes, and the harder it is for others to catch up.
This is how organizations transform how they operate: enabling teams and agents to work from the same, always-current understanding, and to act with speed and consistency across the business.
Start now. Pick a high-impact scenario, connect live signals, and put a single operational loop into production. From there, the system compounds, and so does the lead.
Learn more
To go deeper on how Fabric IQ and Real-Time Intelligence come together, and how to start applying these patterns in your own environment, explore the following resources:
- Build sessions: Real-Time Intelligence | Fabric IQ
- Complete the tutorial: Real-Time Intelligence | Ontology
- Ask questions on the forum: Real-Time Intelligence | IQ
- Submit ideas and vote: Real-Time Intelligence | IQ
- Use the docs: Real-Time Intelligence | IQ
- Complete the learning path: Real-Time Intelligence
- Get certified: Real-Time Intelligence
- Read the blog: Real-Time Intelligence | IQ
- Check the release plan: Real-Time Intelligence | IQ
- Follow on LinkedIn
- Watch on YouTube