The s-X-AIPI project has successfully delivered several results that go significantly beyond the state of the art in industrial AI systems:
Deployment of the s-X-AIPI Autonomic Manager (AM) with validated self-X abilities, including self-optimization. These were implemented using a domain-agnostic reference architecture based on MAPE-K, applicable across multiple industrial settings.
Integration of self-X AI into real-world industrial pipelines, enabling the dynamic adaptation of AI components to unforeseen situations in four use cases (steel, asphalt, pharma, aluminium). The AM coordinates decision-making, monitors system behavior, and triggers corrective actions without manual intervention.
Execution of full pilot-scale validations, demonstrating how AI systems can autonomously optimize quality, efficiency, or circularity. For example:
Predictive adjustments to steel production processes (resilient high-end quality).
Asphalt formulation adapting to recycled material composition in real-time.
AI-assisted root-cause analysis and parameter prediction in pharma.
Automated mix design and sorting logic for recycled aluminium.
Formalization and release of the Reference Architecture in CWA 18211:2025, supporting replicability and transferability beyond the project. This standard sets a benchmark for the integration of AI pipelines with self-management capabilities in industrial environments.
Open-source release of key technological components via GitHub, promoting uptake by the broader innovation ecosystem. Components include orchestration pipelines, metadata layers, self-X function blocks, and modular connectors to legacy OT/IT systems.
Maturity Model and ALTAI-based Trustworthiness Assessment Tools, adapted to industrial users, guiding companies in evaluating their readiness and progress in deploying AI responsibly.
Exploitation and business readiness: Key Exploitable Results (KERs) were identified and assessed for market potential using the AIDA model and validated with the Horizon Results Booster. Early adopter interest was registered in sectors beyond the project pilots.
Scientific and standardisation impact: Contributions to CEN/CENELEC and clustering with sister Horizon Europe projects reinforced the project’s external visibility and influence.
These results set the foundation for more autonomous, adaptive, and human-aware AI systems in Europe’s industrial future, fostering innovation with safety, trust, and replicability at its core.
