How an Energy SaaS Provider Reduced Technical Debt and Became Cloud-Native
Transforming enterprise energy management through real-time IoT streaming, first AI-powered optimization in the early 2020s, and cloud-native infrastructure — delivering measurable cost reductions across large facility portfolios.
At a Glance
| Industry | Energy |
| Client Type | Enterprise energy management SaaS provider |
| Team Size | 6 engineers + 1 ML specialist + 1 cloud architect |
| Engagement | Development partnership |
| Core Stack | Azure cloud-native, .NET 8, React, Azure IoT Hub, Azure OpenAI, Microsoft Fabric |
| Key Outcome | Up to 31% reduction in energy costs; sub-second data latency |
The Challenge
When First Line Software first engaged, the platform faced compounding technical debt: a monolithic ASP.NET application with polling-based data collection every 15 minutes, a fragmented device control architecture requiring per-device management, and a Microsoft SQL Server instance struggling under growing data volumes. Reports were slow to generate, the database was oversized and poorly indexed, and the system could not scale to meet the client’s expanding customer base.
Beyond performance, the client faced a strategic challenge: the energy management market was moving rapidly toward AI-driven recommendations and real-time decision support. Their platform needed to evolve from reactive monitoring to proactive intelligence — or risk losing ground to cloud-native competitors.
What We Built
Cloud-Native Infrastructure on Azure
The platform was re-architected onto Microsoft Azure using a microservices model. The monolithic application was decomposed into independently deployable services — data ingestion, device control, analytics, and alerting — each containerized with Docker and orchestrated via Azure Kubernetes Service (AKS). This enables the platform to scale individual components on demand, reducing infrastructure costs while handling peak loads across hundreds of concurrent enterprise clients.
Real-Time IoT Streaming
The 15-minute polling architecture was replaced with a real-time streaming pipeline using Azure IoT Hub and Azure Event Hubs. Building sensors and management equipment now stream data continuously, with the platform ingesting and processing readings in under one second. This shift from batch to stream processing was foundational — enabling meaningful real-time alerting, control responsiveness, and AI inference on live data.
Unified Device Management
The First Line Software team redesigned the device control layer into a unified command & control service, enabling operators to manage grouped device fleets — HVAC zones, lighting circuits, ventilation arrays — through a single interface and API. The service supports policy-based scheduling, remote override, and emergency shutoff across thousands of devices simultaneously, replacing a device-by-device control model that could not scale.
AI-Powered Energy Optimization
The most significant capability uplift came from embedding AI throughout the platform. Working with Azure OpenAI Service and custom ML models deployed on Azure Machine Learning, First Line Software introduced:
- Predictive anomaly detection that identifies abnormal consumption patterns hours before they escalate, with natural language explanations generated for facility managers.
- AI scheduling recommendations that analyze occupancy patterns, weather forecasts, utility tariff structures, and historical consumption to suggest optimal equipment schedules — surfaced as plain-English recommendations operators can approve or override.
- Automated root-cause analysis that correlates sensor data, maintenance logs, and external factors to explain why consumption spikes occurred and what actions to take.
- Carbon accounting integration that maps energy consumption to real-time grid carbon intensity data, enabling clients to report against net-zero commitments with accuracy.
Microsoft Fabric Analytics Layer
The analytics backend was migrated from legacy SQL Server reporting to Microsoft Fabric, consolidating data engineering, warehousing, and Power BI reporting into a unified analytical platform. This eliminated data pipeline complexity, reduced reporting latency from hours to minutes, and gave the client’s data team a governed workspace to build and share portfolio-level dashboards with their customers.
Modern Frontend Experience
The Building View console was rebuilt in React with a responsive design, replacing the legacy web interface. The new interface provides real-time energy dashboards, AI recommendation cards, interactive floor-plan overlays showing live device states, and mobile-optimized views for facilities staff on the go. Role-based access control ensures enterprise clients can delegate visibility and control across their teams securely.
Results
Since the re-platformed solution went into production, the client has seen measurable impact across their customer base:
Up to 31% reduction in energy costs reported by enterprise customers using AI scheduling recommendations
Sub-second data latency replacing 15-minute polling cycles, enabling real-time operational decisions.
87% reduction in critical energy alerts going unaddressed, due to AI-prioritized alerting and plain-language context.
40% reduction in database infrastructure costs following migration to Azure-managed services and data tiering.
Platform now supports 3x the number of concurrent enterprise clients without additional infrastructure overhead.
Carbon reporting capability enabled customers to meet new mandatory ESG disclosure requirements ahead of regulatory deadlines.
Technologies
| Cloud & Infrastructure | Microsoft Azure, AKS (Kubernetes), Docker, Azure DevOps CI/CD |
| IoT & Streaming | Azure IoT Hub, Azure Event Hubs, Azure Stream Analytics |
| AI & Machine Learning | Azure OpenAI Service, Azure Machine Learning, custom anomaly detection models |
| Data & Analytics | Microsoft Fabric, Azure Data Lake, Power BI Embedded |
| Backend | .NET 8 / ASP.NET Core, microservices architecture, REST & gRPC APIs |
| Frontend | React, TypeScript, Azure AD B2C (auth) |
| Database | Azure SQL, Azure Cosmos DB (time-series device data) |
Updated: May 2026