On the occasion of the General Assembly of the Association of Private Social Security Institutions (AdEPP), dedicated to the world of self-employed professionals in Italy and chaired by President Alberto Oliveti, Aldo Bisio, CEO of Engineering, highlighted the strategic role of Sovereign Artificial Intelligence for businesses and public administration, based on direct data control, process transparency, and economic and energy sustainability.

Bisio stated: “With our IS-IA architecture, we have developed a secure and agnostic infrastructure capable of addressing three fundamental challenges for businesses and institutions. The first concerns the protection of cognitive assets: entrusting processes, data, and distinctive know-how to external models on a rental basis entails the risk of diluting corporate assets. Sovereign AI, on the other hand, allows these resources to be segregated and compartmentalized, protecting competitive advantage and citizens’ data. The second concerns transparency, because as AI evolves to act and make decisions, there is an obligation to explain to citizens, clients, and patients the rationale behind those choices. Unlike ‘frontier’ models with billions of parameters, which operate as closed systems where cognitive biases or the weight of training data are unknown, sovereign AI makes it possible to ‘open the hood’ and understand how logical and semantic processes actually work. Lastly, there are issues related to proportionality, energy savings, and cost reduction. While large global models consume vast amounts of energy, Engineering’s approach focuses on proportionality: using smaller models, with fewer parameters, to achieve the same results for specific tasks. This approach reduces energy consumption and significantly lowers costs associated with ‘tokens,’ which are becoming an increasing expense for companies. Our solution is based on intelligent orchestration capabilities. The goal is to route requests according to proportionality criteria: using large generalist models when data does not require a high level of confidentiality, and relying on proprietary, fenced models to manage sensitive and distinctive information under strict governance control.”

Aldo Bisio spoke alongside Flavio Cattaneo, CEO of Enel; Regina Corradini D’Arienzo, CEO of SIMEST – CDP Group and Stefano Donnarumma, CEO of the FS Italiane Group.