Case study

    Turning legacy uncertainty into modernization clarity with AI‑first delivery

    A customer in the industrial services sector needed a unified portal that would give both internal teams and end‑customers easy access to sensor and telemetry data. Existing analytics solutions didn’t fit their needs: they either lacked the required capability or couldn’t be integrated into their platform.

    Service

    Data and AI

    Industry

    Industry Construction

    A unified portal matters because both technicians and end‑users depend on real‑time data to understand what’s happening and intervene before small issues turn into bigger problems. Without a consolidated trend view, faults surface only after alarms, complaints, or manual inspections, which slows down troubleshooting and drives up operational costs. When telemetry is brought together in one place and visualized clearly, teams can spot changes as they happen, identify anomalies early, and make decisions based on actual building behavior rather than assumptions. The result is more efficient operations, lower energy use, and a more predictable way to manage modern buildings.

    To de‑risk the upcoming modernization program, we ran a six‑week AI‑first proof of concept (PoC) project. It demonstrated the available technical capabilities and showed how our specification‑driven development ensures high‑quality results from AI‑enabled delivery.

    The project delivered a fully functional time-series visualization and analytics application that operates on the customer’s big data stack, integrates with their identity management system, and meets their performance requirements. The PoC built confidence by removing key uncertainties, delivered a working analytics application aligned with the modernization roadmap, and demonstrated over 50% efficiency gains that improve cost predictability for future phases.

    The need for evidence before committing to modernization

    The customer collects large volumes of sensor data, several million data points per day, from buildings, industrial sites, and equipment to their big data platform. They offer monitoring and maintenance services and wanted to enhance their customer portal with deeper data visibility: trend views, multi-sensor comparisons, anomaly exploration, time‑range drill‑downs, and the ability to share direct shortcuts to specific insights with colleagues or customers.

    Despite reviewing existing tools on the market, they found no solution that could both embed seamlessly into their customer portal and align with their user management, branding, and long‑term data architecture requirements. They also needed a platform component that would eventually become part of a broader modernization initiative. The risk was significant: investing in a fullscale capability without knowing if the required performance and scalability were achievable within the budget.

    The customer wanted evidence before committing. A rapid PoC would show whether an AI-first build could deliver the required functionality, integrate with their architecture, and reduce the cost and complexity of modernization.

    A predictable and governed way to build with AI

    We applied our AI‑First Delivery methodology – a structured and specification‑driven approach that blends human expertise with rapid execution done by an agentic AI swarm.

    The team collaborated with the customer to develop formal specifications, using that as the foundation for AI agent workflows. The specification ensures predictable cycles of code generation, review, and testing. In the agentic AI swarm, multiple AI agents were used, and they followed defined roles for code generation, review, correction, and test creation based on our AI‑Kit patterns. Human engineers guided the work by providing architecture and UI design, validated outputs, and ensured quality throughout.

    Our specification‑driven setup ensures that AI work is guided by clear business and technical requirements, executed in controlled generate‑verify‑refine loops, and always governed by human architectural oversight.

    This approach enabled parallel work across frontend, backend, design, documentation, and testing. It also allowed the team to rapidly explore several technical paths and select the one best aligned with the customer’s modernization plans.

    A working analytics application delivered in weeks

    In just a few weeks, the client received a functional application that visualizes multi-sensor timeseries data through an interactive interface. Users can search and select sensors, compare them across multiple axes, zoom and pan through large datasets, review alarm events, and are able to share direct shortcuts to specific insights with colleagues or customers. The application runs on the customer’s technology stack and integrates seamlessly with their authentication flow.

    The solution reads data directly from their Azure Databricks big data environment and processes datasets containing several billion data points. The frontend, backend, and infrastructure were built to be modular, allowing the capability to be extended into a full product and seamlessly integrated into the upcoming portal modernization.

    The customer was able to quickly deploy the solution into their own production environment using the provided IaC scripts. In addition, they received the full source code, documentation, and a demo environment to support understanding and further development.

    Customer was really impressed with the scale of solution and quality of deliverables team was able to deliver within just few weeks.

    Accelerating future modernization

    The solution delivered complete technical capability within weeks and offered several direct business benefits. The structured AI‑First Delivery methodology ensured repeatability, quality, and alignment with the customer’s architecture from start to deployment.

    It reduced the risk for major modernization. By validating performance, scalability, and architectural choices upfront, the customer gained clarity on the feasibility of their long-term program. Beyond demonstrating the analytics module itself, the PoC validated the broader approach. With our AI‑First Delivery methodology and AI‑Kit, custom analytics capabilities can be implemented quickly and consistently - not only in this PoC but also during maintenance and future development.

    It created new capability without long delivery cycles. The customer now has a functioning analytics component capable of serving real users. It delivers new value to their service business by enabling technicians and specialists to see data trends, troubleshoot issues, and understand patterns in building or process behavior.

    It accelerated delivery and reduced costs. AI‑assisted development produced the working solution significantly faster than a traditional approach, cutting the estimated delivery effort by more than 50%. By grounding that reduction in a concrete, working build, the PoC also provided a reliable reference point for estimating the effort needed for future modernization work. That gives the customer a predictable cost baseline for the future modernization program.

    It supports scaling and future offerings. Because the component is part of their architecture and runs on their cloud infrastructure and data platform, it can be extended to new use cases, device types, and a broader set of customers. It also establishes a pattern for using AI‑first development across other areas of the modernization roadmap.

    Get in touch

    Let us offer you a new perspective.