Digital services are undergoing a profound transformation as applications increasingly operate across heterogeneous and geographically dispersed computing environments. These environments include cloud platforms, large-scale data centers, edge clusters, and micro-resources embedded in devices at the far edge. Although this emerging landscape offers flexibility and proximity to data sources, it remains deeply fragmented, creating a barrier to the seamless deployment of highly dynamic and data-intensive applications. This fragmentation is one of the central challenges that ENACT seeks to address within the path toward a true Cognitive Computing Continuum.
Today’s cloud-edge orchestration technologies were not conceived to manage the level of diversity and scale characterizing new hyper-distributed applications. Many solutions are based on static models, manual configuration, or brittle integration mechanisms that cannot react efficiently to changes in network conditions, resource availability, or application demands. At the same time, the growing use of AI-powered services, real-time analytics, and cross-domain data flows requires a fundamentally different approach to infrastructure management, one that is capable of incorporating predictive intelligence, autonomous adaptation, and continuous optimization across distributed nodes.
The compute continuum also faces operational constraints linked to latency, bandwidth, energy consumption, privacy, and location-specific data requirements. These constraints vary significantly across resources, making it difficult for conventional orchestrators to make decisions that reflect the real-time state of the system. Data must be processed where it is most effective to do so, whether in a central cloud environment that offers high computational power or at the edge and far-edge where immediacy and contextual awareness are essential. Meeting these demands requires moving beyond traditional paradigms and adopting dynamic, AI-driven models that can analyze, predict, and coordinate distributed resources with minimal human intervention.
This is where ENACT introduces a paradigm shift. By focusing on dynamic graph-based modelling of edge-to-cloud resources, real-time telemetry-driven adaptation, and multi-source data fusion, ENACT provides a foundation for representing distributed infrastructures as a living, evolving system. The project’s modelling environment captures both functional and non-functional properties of nodes across the continuum, offering insights into connectivity, resource utilization, energy consumption, and even the evolving relationships between services and the underlying physical or logical infrastructure. This holistic view enables application orchestration to move from reactive to predictive, turning infrastructure management into a process driven by anticipation rather than response.
ENACT also incorporates AI Graph Neural Network models for orchestrating data-intensive workflows and managing dependencies between distributed tasks. These models allow the orchestrator to infer optimal scheduling decisions, evaluate the impact of resource constraints, and autonomously adjust application deployments. In addition, ENACT integrates Deep Reinforcement Learning techniques and Kubernetes-based adaptation mechanisms, enabling applications to dynamically rebalance workloads across nodes while maintaining performance, energy efficiency, and quality of experience. Such innovations make it possible to support demanding use cases that rely on immersive media, autonomous systems, and multi-sector data spaces, where applications must continuously adjust to shifting conditions.
By delivering this new generation of intelligent orchestration and adaptive management tools, ENACT confronts one of Europe’s most pressing technological challenges: the need to unify fragmented infrastructures into a coherent, intelligent, and resilient system capable of supporting next-generation digital services. Through its integrated approach, ENACT brings Europe closer to a future where distributed environments can operate as a single, coordinated ecosystem—one that is predictive, autonomous, and inherently aligned with the demands of Web3, AI-driven workloads, and real-time hyper-connected applications.