The grid keeps getting smarter, with AI.
We started in 2012 by reading meters remotely. Today, models learning on the same data find theft, forecast demand, watch switchboards and work as agents inside the platforms. AI is not a feature for us; it is the layer under every platform.
- 01SEIDRA
Theft detection and scoring
A three-layer ML architecture (XGBoost, LSTM, rule engine) yields an explainable theft probability score per subscriber. SHAP-based attribution; more than 3× inspection hit rate for field crews.
- 02HEIM · FORSETA
Demand forecasting and energy balance
Portfolio-level demand forecasting with weather-correlated models, gap filling and P10 / P50 / P90 confidence bands across 21 distribution regions.
- 03VIDOR
Switchboard monitoring and predictive maintenance
Real-time anomaly detection on more than 10,000 low-voltage switchboards with LSTM-AE and Isolation Forest models, automatic alarms and operator recommendations.
- 04HUGYN
Market forecasting
Tomorrow's hourly PTF, SMF and system direction for the rest of today; every forecast is scored at 14:00, the daily brief ships before the day-ahead gate and refreshes live as data changes.
- 05LENA
Agents in the metering infrastructure
An AI-assisted fault-analysis engine, user support agents and field analytics; the lifecycle of millions of meters managed at national scale.
- 06MYMIR
Enterprise data fabric
An AI Data Fabric that discovers scattered enterprise databases with LLMs, builds a knowledge graph and gives non-technical users natural-language access to data.
- 07ORIN
One gateway for APIs, LLMs and MCP
The API gateway and ESB that runs the organisation's application-to-application and external integrations routes AI traffic through the same gateway: token and cost metering on local and cloud LLM traffic, RBAC-scoped access to MCP servers, every tool call monitored like an API call, and all of it in one SIEM-ready audit trail.
