Blog | Microsoft FABRIC
Practical AI:
Empowering Users with the Microsoft Fabric Data Agent
19 June 2025
Jack Mysyk & Elizabeth Phillips
Data and Analytics team at Decision Inc. UK
Based on forecasts by Gartner, Generative AI (GenAI) spending is expected to total $644 billion in 2025, marking a 76.4% increase from 2024. As adoption accelerates, businesses are actively exploring how GenAI can drive real operational value. However, many organisations still face uncertainty around how to apply GenAI effectively—particularly in ways that benefit both technical and non-technical users.
A longstanding Decision Inc. client had a question many organisations are starting to ask:
How can GenAI support our day-to-day operations?
The Challenge: Insight Bottlenecks in AP
Their Accounts Payable (AP) team wanted to focus on improving how they interact with data. Currently, the team relies on a Power BI semantic model and report to monitor and analyse outstanding payables. While the model is thorough and well-designed, navigating it to extract specific insights can be time-consuming—especially for users who are not as familiar with the data structure.
The Opportunity: GenAI + Microsoft Fabric
To bridge this gap between data and decision making, Decision Inc. developed a proof of concept (PoC) leveraging Microsoft Fabric’s Data Agent. This feature, launched by Microsoft in 2025, enables the creation of customisable Q&A interfaces powered by GenAI—making advanced data insights directly accessible across the business, regardless of technical expertise
We set out with a clear path forward:
Leverage the client’s existing Microsoft Fabric investment
Build a natural language interface tailored to the AP Team’s needs
Maintain existing security and governance standards
Our Solution: The AP Invoicing Data Agent
Decision Inc. designed and built a GenAI-powered Q&A interface using Microsoft Fabric’s Data Agent that allows the client’s users to ask plain-language questions about their AP data.
What We Did
Migrated the existing semantic model from Power BI to Microsoft Fabric’s warehouse, preserving calculated columns and critical data structures.
Refactored dimensional tables to better align with the client’s business terminology and common AP questions, enabling accurate interpretation of natural language queries.
Created a custom Data Agent artifact within Fabric to serve as the interface for GenAI-powered querying.
How It Works
The Data Agent reads data directly from the Fabric warehouse in real-time, minimising any latency between semantic model refreshes.
It was trained with 15+ example queries, such as: “Show me all open invoices associated with Purchase Order X.” This allows the GenAI model to quickly comprehend and respond to frequent user requests.
Business context prompts and synonym mappings were embedded to bridge the gap between business users and required datapoints.
The Benefits
The AP Invoicing Data Agent PoC delivered immediate and measurable improvements to how the client’s AP team interacts with their data.
Key outcomes included:
Faster Insight Generation
Users can now ask questions using natural language queries and receive quick answers, saving time and reducing reliance on analysts.
Tailored to Business Users
The Data Agent was trained with real-world examples and aligned with the client’s terminology, ensuring accurate and relevant responses..
Scalable by Design
The solution is future-ready, allowing the integration of additional tables and data sources as the AP team’s requirements grow.
Secure by Design
The Data Agent honours user permissions and implements Row-Level Security (RLS) and Column-Level Security (CLS).
Integrated into Daily Workflows
The Data Agent is accessible directly through Microsoft Teams, which makes insights available within the tools users already rely on
Looking Ahead
The implementation of Microsoft Fabric Data Agents is a significant step towards creating a data-driven culture—where insights are accessible to everyone, not just analysts. By facilitating natural language conversations with data stored in Fabric’s OneLake, even non-technical users can derive meaningful insights effortlessly.
This case demonstrates how GenAI can move beyond hype and into real-world utility by providing intuitive, self-service insights on a unified, scalable platform.
Interested in discovering how GenAI can bring practical value to your business functions and deliver real impact? Get in touch with our team today to explore tailored solutions designed to meet your unique needs. Speak to an Expert