The European Commission has reinforced the central role of data in artificial intelligence through two major policy initiatives in 2025. The AI Continent Action Plan, published in April, underlines within its “Data for AI” pillar that access to reliable and well-organised data is essential for the EU to unlock the full potential of AI. This message was further developed in the European Data Union Strategy released in November, subtitled Unlocking data for AI, which sets out the need to scale up access to high-quality data to drive AI and innovation.
“Data for AI” has long been a priority area for the BDVA community and formed a core element of its responses to these European Commission initiatives. These contributions were published in the papers Towards a European AI-Data Value Ecosystem and Data at the core of Europe’s Digital Strategy.
Building on this work, the BDVA community has identified an urgent need to evolve the traditional concept of data products. While existing data products focus largely on packaging and sharing datasets for general use, they are not sufficient to meet the specialised requirements of AI systems. BDVA therefore introduces a new paradigm: AI-ready Data Products.
A newly published paper consolidates past and ongoing discussions within the community on how the “Data for AI” paradigm can be embedded into data product approaches. It sets out a pathway for redefining data products so that AI practitioners can fully harness the value of data within increasingly complex data ecosystems. By doing so, it aims to position these ecosystems as strong enablers of AI innovation, capable of delivering industry-ready solutions tailored to advanced AI applications.
According to BDVA, this represents the first systematic attempt to adapt the data product concept to the specific needs, constraints and practices of AI development. The work identifies several key requirements. These include revisiting the data product lifecycle to incorporate AI-specific elements, particularly richer and more extended metadata; addressing technical challenges related to data quality, preparation techniques and metadata models suited for AI consumption; and supporting both static and dynamic data products aligned with AI workflows.
The paper also highlights the need to strengthen governance and compliance frameworks, including ethical considerations, regulatory alignment, standards, and licensing and contractual aspects relevant to AI models and AI-ready data products. To support adoption, the community calls for a readiness framework that captures these broader requirements and provides practical guidance.
The convergence of data management best practices, modern data product principles and AI-specific needs is seen as a highly promising route towards truly AI-ready data products. BDVA emphasises that further exploration, cross-disciplinary collaboration and validation are required. The community plans to continue this work internally, including the release of an updated version of the paper in 2026, while also expanding the dialogue with external stakeholders such as other associations, standardisation bodies, policymakers, industry actors and AI communities.
Ultimately, BDVA aims to translate these initial concepts into concrete standards, tools and methodologies that can be deployed across sectors. By doing so, the community seeks to contribute to the successful implementation of the European Data Union Strategy and to the creation of an integrated European AI–data value ecosystem that effectively links infrastructure, data, AI, skills, regulation and innovation.