Agro-Sieć is an authorized John Deere dealer and Poland's most recognized company in the agricultural machinery sector. Operating an extensive network of dealerships and service centers, the company sells, rents, and services machines that farmers across the country rely on every season. Among Agro-Sieć's clients are professionals managing large agricultural operations, with precise expectations around product quality, after-sales service, and account transparency.
As their customer base and product portfolio grew, Agro-Sieć needed digital solutions that would support their operations and improve the experience they deliver. That ambition required something most growing companies eventually face: a reliable, unified view of their own data.
The Challenge
Agro-Sieć kept their data in multiple systems, such as ERP, CRM, and hundreds of Excel files. These platforms held valuable but disconnected information. It resulted in:
The platform also needed to account for a mid-term integration of IoT sensor data from tractors, with the expectation that these metrics would become a key input for predictive maintenance, operational optimization, and more advanced planning use cases. The company had already taken a first step by adopting Microsoft Fabric, but its usage had stayed limited: simple notebooks without a scalable architecture or transformation pipeline underneath. There was no clear data model, no automated ingestion, and no clean layer that analytics tools could confidently read from.
As Microsoft Fabric was a relatively new product, it was still carrying early-stage limitations, evolving features, and occasional instability or performance limitations in some components that the team could not fully anticipate from the start.
Building on top of an evolving platform meant navigating challenges that went beyond what documentation alone could fully address. To overcome the challenges of fragmented data and the early-stage maturity of the platform, Agro-Sieć needed a partner who could bring both deep data engineering expertise and the technical resilience to work through those unknowns with them. And that's how STX Next entered the picture.
The Solution
Our engagement began with a thorough assessment of what Agro-Sieć had and what they needed. The team defined the architecture, then set up and configured the platform and delivery stack end-to-end: infrastructure managed through Terraform where applicable, CI/CD pipelines via GitHub Actions integrated with the client's existing repositories, and a structured Microsoft Fabric environment designed to grow as requirements evolve.
The data architecture follows the Medallion model: raw data lands in a Bronze layer, gets cleaned and standardized in a Silver layer, and surfaces as analytics-ready models in the Gold serving layer.
This separation makes the pipeline maintainable and supports more isolated testing and validation per layer, which matters enormously when requirements change or new data sources are added.
It also allows refresh rates for individual reports to become a business decision rather than a technical constraint, adjusted through configuration to balance data freshness with compute cost depending on the use case.
The platform we built for Agro-Sieć on Microsoft Fabric combines Data Lakehouse and Data Warehouse capabilities available within the platform, each serving a distinct role on top of shared OneLake storage.
The Data Lakehouse collects and stores raw data and implements the medallion architecture, where data is incrementally refined across layers.
The Data Warehouse builds on top of curated data to provide a structured, SQL-based serving layer optimized for analytics and reporting. The Data Lakehouse serves as a common entry point for data from various systems, while both components remain logically separated but operate on the same underlying storage.
Within the Warehouse, data is organized into layers (bronze, silver, gold) to ensure it is progressively cleaned, standardized, and made analytics-ready. This setup supports scalable ingestion and transformation pipelines. This approach was driven by the fact that the Data Warehouse provided a more natural fit for SQL-based transformations with dbt, while the Data Lakehouse offered greater flexibility for ingestion and raw data processing.
As a result, each component and data layer has a clear role, and ingestion and transformations are properly separated from serving and consumption, avoiding unnecessary coupling across the platform.
For transformations and data modeling, the team chose dbt. It was the right call for this type of work: dbt brings reusability and testing into the transformation layer, with version control handled through Git integration.
The model can be documented, every dependency is explicit, and the transformation pipeline can be reasoned about as code, with dbt's built-in testing, documentation, and lineage making long-term maintenance far easier.
The first phase successfully delivered an end-to-end pipeline for the ingestion and modeling of Agro-Sieć's ERP data. That milestone represented the core of the project: structured, reliable data flowing from source systems through the pipeline into an analytics-ready layer.
The data lakehouse integrates directly with Agro-Sieć's existing ERP and CRM systems, including MS Dynamics on top of MS Dataverse, source of the data.
That connection was essential: the value of the new platform depended on the accuracy and completeness of the data it ingested. Ensuring clean, validated integration with those legacy systems required careful work and close collaboration with the client team.
The Impact
ERP and CRM data is now standardized and queryable from a single serving layer, replacing a fragmented mix of disconnected systems and Excel files that required manual reconciliation to produce any consistent report.
The Medallion architecture and dbt modeling layer are modular by design, so the platform grows by extension rather than by rework. Adding new data sources, extending the serving layer, or building new analytics products does not require rearchitecting what already exists.
Analysts no longer depend on manually assembled data. The Power BI reporting layer connects directly to the Warehouse via Direct Lake, giving business users access to operational metrics without preparing data by hand before each report.
The entire infrastructure is managed as code where applicable. Environments are reproducible, deployments are automated, and the team does not depend on manual configuration steps to maintain or extend the platform.
The architecture supports predictive maintenance and operational optimization use cases without needing a redesign when those data streams are connected.
The first end-to-end pipeline for ERP data ingestion and modeling was delivered in the initial phase, giving the business a working, production-grade data flow from source systems through to the analytics layer from day one of the platform's active use.
The Partnership
Technical success is driven by strong relationships. Agro-Sieć proved to be an open and communicative partner, trusting our judgment while navigating the less-charted aspects of Microsoft Fabric. Through honest communication and a shared focus on business interests, we successfully mitigated risks and turned a complex challenge into a robust data foundation.
Schedule a call with our experts to discover the benefits of partnering with STX Next.
Contact us→Our customers love to work with us
ul. Mostowa 38
61-854 Poznań
Poland
C.11 310 Col. Santa Gertrudis Copo
Merida, 97305
Mexico
10 York Rd
London SE1 7ND
United Kingdom
c/o Rödl RAe, Taunus Tower, Mergenthalerallee 73-75
65760 Eschborn
Germany
City Place 1401 Lake Plaza Dr, Ste. 200
Spring, TX 77389
United States
