• Can you briefly introduce your company and what you do?

XILBI Sistemas de Informacion SL (Spain) is a European IT and innovation company specialised in edge computing, cloud-native software, AI, IoT applications and secure digital systems. XILBI combines expertise in AI/ML, energy systems, DevOps, cybersecurity and data interoperability. The company has a strong track record in European R&D and commercial innovation, including work in renewable energy, Earth Observation, smart agriculture and critical digital infrastructures. In the energy domain, XILBI has developed advanced solutions for forecasting, anomaly detection and decision support, and operates a renewable energy Living Lab in Salamanca, Spain for testing and validation. Through SEDGE, XILBI is bringing this experience into the O-CEI ecosystem to deliver reusable, privacy-preserving synthetic energy data assets for AI and edge applications.

• What challenge are you addressing under the O-CEI Horizon’s first Open Call, and how is your proposal relevant to the challenge?

Under the O-CEI Horizon’s 1st Open Call, we address a core challenge in Cloud-Edge-IoT energy innovation: the lack of privacy-compliant, high-fidelity and interoperable datasets for training, validating and benchmarking AI/ML models. In real energy environments, access to operational data is often limited by privacy, commercial sensitivity, fragmentation and the cost of collecting representative edge scenarios. This makes it difficult to develop robust AI solutions for forecasting, non-intrusive load monitoring, demand response and other edge analytics.

Our proposal, SEDGE – Synthetic Energy Data Generation Engine – responds directly to this challenge by creating a synthetic data engine tailored to energy and edge environments. Within O-CEI Pilot 1.1 in Ireland, SEDGE generates realistic time-series data for residential and small-tertiary scenarios involving Linky smart meters, PV systems, batteries and flexible loads. It combines physics-based simulation with generative AI, behavioural modelling of occupants and a constraint-validation layer so that generated datasets remain both realistic and operationally feasible.

The proposal is especially relevant to O-CEI because it is designed as a reusable CEI asset rather than a one-off demonstrator. SEDGE packages synthetic datasets, edge-ready corpora and pre-training recipes that can be used on Linux gateways and K3s clusters, enabling innovators to test and benchmark solutions before using real customer data. It also encodes outputs with SHIFT ontology semantics, making datasets machine-discoverable, interoperable and ready for publication in the O-CEI marketplace. In this way, SEDGE helps reduce data-access barriers, supports trustworthy AI development at the edge, and contributes reusable assets to the wider O-CEI ecosystem.

• What is the expected impact of your proposal?

The expected impact of SEDGE is technical, operational and market-facing. First, it will provide O-CEI and Pilot 1.1 with a reusable synthetic energy data capability that enables faster and safer development of AI models for edge use cases such as forecasting, NILM and demand response. By generating realistic, privacy-preserving datasets, SEDGE reduces dependence on scarce real-world data and allows solutions to be tested before deployment in operational environments.

Second, SEDGE is expected to improve model quality and development efficiency. Our objective is to demonstrate that models pre-trained on SEDGE datasets and then fine-tuned on real data achieve measurable gains compared with training from scratch, while also reducing edge training time. This can shorten experimentation cycles, reduce field testing effort and help innovators validate solutions more quickly within the Cloud-Edge-IoT continuum.

Third, SEDGE has strong interoperability and replication value. All outputs are prepared with SHIFT-compliant semantics and dataset cards, making them suitable for discovery, reuse and benchmarking through the O-CEI marketplace. Because scenarios are parameterised by weather, tariffs, building archetypes and technology configurations, the approach can be adapted beyond the French pilot to other regions and Member States with limited additional engineering.

Finally, SEDGE contributes to trust and sustainability. It supports privacy-preserving AI development, reduces the need for resource-intensive field data collection, and helps organizations simulate rare or high-risk scenarios without exposing personal or commercially sensitive data. Overall, the project aims to establish synthetic energy data as a practical, trustworthy, and scalable enabler for European edge AI innovation.