- Can you briefly introduce your company and what you do?
Plegma Labs specialises in enterprise IoT, energy analytics, and AI, offering a hardware-agnostic IoT platform, coupled with edge-level modules, unifying heterogeneous meters and sensors, supporting industrial/building protocols, and providing advanced analytics and visualisation for residential, commercial, and industrial environments. Plegma Labs has proven expertise in AI, IoT, and energy analytics, backed by a strong track record in commercial and EU-funded projects (https://pleg.ma/projects/). Our team combines strong scientific backgrounds and peer-reviewed publications (e.g., bit.ly/4e6ixFA, bit.ly/3HzyEiC, bit.ly/3HRnZAD) with hands-on engineering experience, covering AI model development, IoT data integration, and scalable system deployment. Overall, the vision of the company is to enhance the environmental impact and sustainability of companies and citizens through transparent IoT adoption.
- What challenge are you addressing under the O-CEI Horizon’s first Open Call, and how is your proposal relevant to the challenge?
ELEXIR advances the goals of O-CEI by facilitating edge energy insights, privacy-preserving analytics, and interoperable energy services within the Cloud-Edge-IoT continuum. ELEXIR develops and validates an edge-AI framework that transforms household energy data into actionable intelligence for flexible, low-carbon consumption in the context of O-CEI’s Pilot 1.2 (France). Its contribution extends beyond current market practice, where residential energy services mostly rely on cloud-based systems and low-frequency load data, by enabling device-level insights, decision making, and explainable flexibility assessment directly on residential IoT gateways. This adaptation of advanced AI methods to constrained edge environments, while maintaining privacy, distinguishes ELEXIR from the state of the art and reinforces O-CEI’s ambition to catalyse reusable, scalable, cross-domain edge services. ELEXIR ingests fine-grained load curves and adopts deep learning models for Non-Intrusive Load Monitoring (NILM) and short-term forecasting directly on residential gateways. The models quantify device-level usage and flexibility without exposing raw data, ensuring privacy, while Explainable AI (XAI) methodologies will also enhance consumer energy literacy, explaining energy consumption and flexibility potential. In addition, Federated Learning will be used to refine the NILM models across households by sharing model updates rather than raw data, increasing accuracy while maintaining privacy. A Reinforcement Learning (RL) agent offers actionable recommendations for shifting device energy consumption outside peaks and maximizing the utilization of RES production. As far as edge deployment within the O-CEI ecosystem is concerned, deep learning optimization techniques such as quantization and pruning will be adopted, enabling high-frequency inference on constrained hardware, while the developed solution will be accessible via the O-CEI marketplace.
- What is the expected impact of your proposal?
ELEXIR reduces environmental impact by shifting residential electricity demand away from carbon-intensive peaks toward periods of higher renewable availability, cutting both peak-related grid losses and marginal emissions. On-gateway analytics automatically identify household appliances whose operation can be flexible and estimate their shifting potential at the device level. The approach builds on evidence that accurate, device-specific disaggregation and feedback can drive meaningful efficiency gains in homes, providing targeted interventions for major loads without intrusive submetering. Reinforcement Learning controllers have shown significant potential in residential and building energy management, delivering peak-load and cost reductions when coupled with short-term forecasts and comfort constraints. In addition, processing data locally on the residential gateway avoids continuous transfer and storage of raw consumption streams in the cloud, which lowers the indirect energy footprint associated with data-centre activity.
By enabling responsive demand and flexible energy use at scale, ELEXIR contributes directly to the objectives of the European Green Deal, the Fit for 55 Package, and the EU Climate Law, which collectively aim to reduce greenhouse gas emissions by at least 55% by 2030 and achieve climate neutrality by 2050. The project operationalizes these policy goals at the household level, empowering consumers to participate in the green transition through measurable, automated actions and behavioral change via detailed consumption insights.
ELEXIR delivers a social impact by turning citizens from passive energy consumers into active participants in Europe’s green and digital transitions. Through explainable, human-in-the-loop AI, households gain transparency into when, how, and why their energy is used, empowering them to make informed decisions and directly contribute to decarbonisation efforts.
