Finansforbundet builds a data agent for financial management

Finansforbundet wanted to make financial data directly usable for management and decision-makers without requiring deep technical insight. Leaders needed fast, precise answers to concrete business questions such as how much budget remained for the year—without relying on complex systems, static reports, or manual analyses.

Although the organization already had access to relevant data, it lacked an operational layer that could translate that data into actionable insight. At the same time, it was critical that the solution did not become a closed black box. Finansforbundet wanted transparency, governance, and ownership—with the ability to explore AI capabilities on the Microsoft platform and continue developing the solution without long-term dependency on external consultants.

Context& designed and facilitated a structured, outcome-driven delivery model focused on rapid value creation and long-term ownership. The approach combined hands-on development with systematic competence building, ensuring that new capabilities could be applied in daily operations.

The delivery was organized as a series of focused workshops with clear structure and progression. Context& established the architectural foundation, design principles, and concrete implementation examples. Between sessions, Finansforbundet worked independently, applying learnings in practice. Each workshop started with review and sparring before introducing new topics and next steps.

The technical foundation was built using Microsoft Fabric and Copilot Studio, with a clear focus on modular, context-based agents rather than a single generic solution. The work included semantic modeling, metadata design, response optimization, and the establishment of a ‘golden dataset’ to systematically validate agent responses.

In parallel, Context& helped define a clear governance model covering publishing, security, access rights, and lifecycle management. This ensures the solution can be used broadly across the organization and continuously extended as needs evolve.

Finansforbundet now has a working prototype of a Fabric Data Agent that answers financial questions directly based on the organization’s own data.

The internal team has acquired the skills needed to maintain, adapt, and extend the solution independently. A solid foundation for governance, integration, and scaling is in place, reducing dependency on external consultants and shortening decision-making cycles.

The case demonstrates how agentic AI creates real value when technology delivery and competence building go hand in hand—and when ownership is anchored in the organization from day one.