AI adoption requires advanced computing infrastructure, and just as much reliable, interoperable data that is ready to use. This is the subject of the paper “Abilitare l’Ecosistema Digitale con Dati AI Ready: il Modello IT4LIA” (Enabling the Digital Ecosystem with AI-Ready Data: the IT4LIA Model), written by Donatella Sforzini (CINECA) and Arnaud Ceol (Fondazione ICSC, National Research Centre for HPC, Big Data and Quantum Computing) and published among the selected papers of the GARR 2026 Conference “The shape of change – Between digital ethics and shared roots”, held in Pisa on 19-21 May.
The paper describes the approach IT4LIA AI Factory adopts to catalogue, prepare, share and make the most of data within the national digital ecosystem.
A platform for European innovation
The strategic vertical sectors of IT4LIA, where the initiative has specific expertise, are Agritech and Agrifood, Earth, Manufacturing and Cybersecurity, with openness to all sectors, for example Cultural Heritage, Health and Finance.
The infrastructure includes the EuroHPC supercomputer LEONARDO and its upgrade LISA, the GAIA cloud system, MEGARIDE and the new IT4LIA supercomputer, optimised for artificial intelligence. On top of this base come technology services, specialist support and training programmes for SMEs, startups, research institutions, public administrations and public decision-makers. Among the distinctive features of IT4LIA, the authors point to its focus on open science and on making the most of data.
AI-ready data across the whole data lifecycle
One of the main obstacles to AI adoption is the fragmentation and heterogeneity of the information available. The AI Readiness strategy of IT4LIA aims to turn very different collections of data into resources that can be used directly in model training, validation and inference workflows.
The approach follows the entire data lifecycle. Information is acquired from on-premise systems, distributed infrastructures, public repositories or external providers, and then goes through harmonisation, metadata enrichment and validation. The preparation phase includes cleaning, anonymisation, normalisation and, where necessary, the generation of synthetic data.
The National Data Registry
The National Data Registry (Registro Nazionale dei Dati) is the service that enables the whole AI Factory ecosystem: it aggregates, catalogues and makes accessible datasets from heterogeneous sources. Its architecture is based on two levels of metadata. Functional metadata describe meaning, application context, conditions of use, regulatory constraints and governance aspects. Technical metadata document structure, formats, provenance, transformations and quality indicators.
The governance model is oriented towards quality, traceability and regulatory compliance. According to the authors, it supports the application of the FAIR principles and meets the requirements of the European regulatory framework, including the AI Act.
A platform open to the community
In their conclusions, Sforzini and Ceol stress that the sustainability of the model depends decisively on the active participation of the community. IT4LIA is a collaborative platform, open to contributions from research institutions, public administrations, industry and SMEs. Interested parties are invited to take part by proposing innovative projects and by contributing to the National Data Registry, thereby enriching the shared body of information. Strengthening data services and encouraging community participation are, for the authors, the key elements for realising the full potential of artificial intelligence.
Read the paper
The full text is available in the proceedings of the GARR 2026 Conference “The shape of change – Between digital ethics and shared roots” (the paper is in Italian): Abilitare l’Ecosistema Digitale con Dati AI Ready: il Modello IT4LIA.
Find out how IT4LIA AI Factory can support your organisation
