Microsoft Fabric
data agents DEMO
Microsoft Fabric Data Agents let analysts, executives, and operations teams ask questions about enterprise data in plain English — no SQL, DAX, or KQL required.
Each agent is scoped to a specific data domain and acts like a virtual analyst for that area. A user types a question; the agent translates it into the appropriate query language, runs it against the underlying data source, and returns a direct, contextual answer. No dashboards to navigate. No tickets to file with a data team.
The interaction model is conversational and stateless per session. Users type a question, receive an answer grounded in live data, and can follow up naturally. The agent reasons across the data rather than simply retrieving it — meaning it can handle multi-step questions that would otherwise require manual joins or report requests.
Data sources span the full Fabric ecosystem: Lakehouses, Warehouses, Power BI semantic models, KQL databases, mirrored databases, and unstructured content indexed through Azure AI Search (PDFs, documents, contracts). This makes a single agent capable of bridging structured and unstructured knowledge domains.
Governance is inherited, not configured. Access runs under each user's own Microsoft Entra identity. The agent only surfaces data the user already has permission to see. Microsoft Purview sensitivity labels and access policies are enforced automatically — meaning self-service analytics doesn't create new governance exposure.
In multi-agent architectures, individual data agents act as modular, domain-specific nodes. They can be called by orchestrating agents or surface insights into external apps via standard APIs, making them composable building blocks rather than standalone chatbots.
The design challenge is clarifying what the agent can and can't answer — and gracefully handling the edge cases where the question outpaces the data model. Transparency about confidence, scope, and data freshness is where the UX earns or loses user trust.
Microsoft Fabric IQ DEMO
Microsoft Fabric IQ is the semantic intelligence layer that sits between raw enterprise data and the people and AI agents who need to act on it. Announced at Microsoft Ignite 2025, it represents a shift in how Fabric is positioned — not just as a place to store and query data, but as a system that understands what that data means in business terms.
The core problem it solves
Organizations are drowning in data without understanding the semantics. An airline doesn't think in terms of tables and schedules — it thinks in terms of flights, passengers, and delays. But that meaning lives in people's minds. Teams use their own definitions and reports. AI systems can read data, but they lack the business context to make reliable decisions.
Fabric IQ is the layer that closes that gap.
What it is
Fabric IQ is the new semantic intelligence layer that elevates Fabric from a unified data platform to a unified intelligence platform — enabling every team and every AI agent to reason, decide, and act in the language of the business, powered by a live, unified view of the enterprise.
It brings three layers of business context together: unified data, business intelligence, and operational intelligence — delivered through two core items in the Fabric IQ workload: ontologies and semantic models.
The ontology — the key structural concept
At the core of Fabric IQ is the Ontology item. It introduces the semantic foundation that connects people, processes, systems, actions, rules, and data into a unified model. By binding real-world data to these ontologies, raw tables and events are elevated into rich business entities and relationships — giving people and AI a higher-level, structured view of the business to think, reason, and act with confidence.
Practically, this means concepts like "Customer," "Shipment," or "Cold Chain Breach" are defined once and reused consistently across reports, agents, and applications — rather than being redefined independently by every team.
How it fits into Microsoft's broader intelligence stack
Fabric IQ is part of Microsoft IQ, the enterprise intelligence layer of the Microsoft stack. It works alongside Work IQ (context on how employees work), Foundry IQ (context from policies and authoritative documents), and Web IQ (context from the web) — with Fabric IQ providing context on business entities and data.
Real-time and agentic
Together, Real-Time Intelligence and Fabric IQ create a unified real-time decision system. Real-Time Intelligence senses what is happening in the moment; IQ interprets it and drives the next action — enabling enterprises and their AI agents to operate with precision even as conditions change second by second.
The design challenge
The UX problem Fabric IQ introduces is one of abstraction visibility. Users interacting with agents or dashboards powered by Fabric IQ are reasoning against a semantic model they likely can't see or modify. The design work lives in making that model legible — surfacing which business concepts are defined, how they're related, and where the definitions came from — without overwhelming the experience with data architecture. Trust in AI-generated answers depends heavily on users understanding, at least at a high level, that there's a governed semantic foundation underneath.