In recent years, big data has generated vast amounts of information, often heterogeneous, with misaligned semantics and difficult to integrate.

What we need instead is high-quality data (better data). Today, data loses its value very quickly if we cannot transform it rapidly into reliable, usable information. This effect is amplified in healthcare organisations: processes are complex, demand is growing, constraints are stringent, and information systems have often evolved in layers over time.

Without trust and timeliness, AI - and GenAI in particular - risks becoming yet another source of complexity.

So how can we trust data and ensure we can use it at the right time?

Every minute spent searching for data is a minute taken away from decision-making, process improvement and clinical support. Looking at local health authorities (Aziende Sanitarie Locali, or ASLs) in practical terms, the most common challenges fall into three main areas:

We need to move away from a project-based approach towards a data product model, treating data as a managed, reliable asset.

For a local health authority, a data product means having data packages designed for specific use cases and levels of consumption, governed by clear rules covering shared definitions, quality, traceability and lineage, access and privacy policies, and explicit service levels.

The direction is towards a more advanced organisational model: a federated one. The solution is neither to leave everything to individual projects nor to place everything under a single central governance body, which can create similar delays.

The key elements are a data platform (not merely an application platform, but one that supports data integration, modelling and access) combined with central standards and accountable domains.

These domains, covering areas such as hospital admissions, emergency departments, specialist care, integrated home care, pharmaceuticals and waiting lists, build and maintain their own data products in parallel, following shared rules.

Trust emerges when ownership and rules are clear, and when data comes with all the necessary information and guarantees. With data products, trust is built because:

This fundamentally changes the relationship between data producers and consumers: a data product becomes an offering with measurable service commitments.

For local health authorities, the benefit is tangible: less time spent reconciling data and more time devoted to analysis and decision-making.

How can we make an AI assistant reliable, verifiable and useful?

To unlock the full potential of GenAI, we need to retrieve certified context from enterprise data and generate responses that can be verified and traced back to their sources. This can be achieved through approaches such as retrieval and grounding, which anchor responses in reliable data.

For these systems to work effectively, they require clean, consistent and well-governed data - precisely what data products enable.

The real ROI lies in reducing operational friction, repetitive tasks, emails, tickets and manual data extractions - ultimately freeing up capacity.

Data quality is the means by which we enable healthcare organisations to make better decisions, respond more effectively, plan ahead and continuously improve.
AI is not the starting point: it is the destination of a journey in which data becomes a product, a service and a shared asset.