Most governments already have artificial intelligence (AI) governance mandates in place: nearly all OECD countries have an AI strategy, and 83% have designated the institutions responsible for overseeing it. These high-level strategies set out objectives and roadmaps for adopting AI in government responsibly, efficiently, and coherently.

Under this broader mandate, many public institutions have initially approached AI through experimentation in controlled, isolated environments. This approach follows a reasonable logic of responsible innovation: identify clear use cases, test a solution’s viability in a test environment and - if the pilot succeeds - scale it to deliver real-world impact. However, scaling is far from straightforward, as we have previously explored in Gobe Insights.

While experimenting before committing resources is good practice, a problem arises when the proliferation of pilots leads different teams within the same organisation to reinvent the wheel every time a new opportunity for AI emerges, solving identical problems separately. Developing AI solutions in silos creates two major issues. On the one hand, duplicated efforts cause inefficiencies and drive up unnecessary costs. On the other, fragmented initiatives make it harder to ensure technical coherence, manage risks, and maintain shared service quality standards.

The case for reuse

To tackle this challenge, an increasing number of institutions are laying the groundwork to make sharing and reusing AI tools and components the most agile, effective path to adoption across users, teams and agencies. At the European level, the Interoperable Europe Act establishes governance structures designed to promote inter-administrative cooperation. One example is the creation of the catalogue of interoperability solutions, which allows European public administrations to propose and discover reusable technology solutions. At the regional level, the Government of Castilla-La Mancha recently launched a new corporate AI framework that explicitly encourages solution reuse to ensure technological consistency across its ministries.

In practice, scaling and sharing AI solutions across an entire organisation requires more than just shared governance infrastructure and data-sharing systems. It demands a robust governance framework, public employees equipped with the skills to leverage the available tools, and long-term ownership and operating models capable of sustaining the development and maintenance of reusable solutions over time. 

The CTTI case: how to unlock solution reuse in public administration

One public institution taking proactive steps toward expanding the availability of its AI solutions across the organisation is the Centre de Telecomunicacions i Tecnologies de la Informació (CTTI) of the Government of Catalonia. To date, CTTI - alongside Catalan government departments and technology providers - has developed hundreds of AI pilots to streamline internal operations and improve public services. Several of these are already in production and offer varying levels of potential for broader reuse. CTTI’s strategic objective is now to deliver corporate technology services under a unified governance framework for AI project development.

To achieve this, CTTI provides civil servants and suppliers with advanced analytics tools to drive Catalonia’s digital transformation in a standardised, secure and well-governed manner. The organisation also ensures public accountability through its public register of cross-cutting services. A working example of an operational cross-cutting service is GD. Anonimitzador, co-developed with Nymiz, which uses AI to identify and irreversibly redact sensitive data in documents and images.

At Gobe, we are supporting CTTI in defining its strategy for identifying and deploying reusable AI components and solutions. Our starting point is acknowledging the existence of multiple projects, spread across different departments, tackling the same underlying challenge - a clear indicator that a cross-cutting solution may be needed. A representative example is voice-assisted AI for form filling. This technology is applied across multiple use cases to address a shared, recurring need: recording data reliably and consistently without increasing the administrative burden on public servants or citizens.

To guide this strategy, we categorise CTTI’s AI initiatives into functional areas, identifying the primary AI capability driving each project - ranging from internal document search assistants to automated data extraction and classification. This functional prioritisation is a critical first step aimed at promoting solutions where AI delivers high value to a large volume of users across different contexts. In parallel, a technical assessment should determine which tools are best suited for cross-departamental rollout based on their alignment with reference architectures, scalability, and modular design.

While these two layers of analysis help decide which solutions belong in the corporate service catalogue, simply making a component available is not enough. Effective adoption requires every deployment to be paired with clear user documentation and well-defined roles and responsibilities, including maintenance, updates, quality assurance, incident resolution and cost-allocation models among adopting departments.

Sharing lessons learned as a first step

The CTTI case illustrates a viable roadmap, though it requires a level of AI maturity that many public administrations have yet to reach. For organisations where teams are still duplicating efforts as they navigate AI adoption, the first step toward breaking down silos may be to increase the visibility of existing projects.

The UK government - widely recognised as a top performer government in AI readiness - promotes this knowledge sharing through the AI Knowledge Hub. The portal aggregates tools built across the public sector, outlines their specific use cases and outcomes, and provides contact details so teams can adopt existing solutions or exchange insights with the original developers. While a platform like this lays the foundation for reuse by gauging interest in solutions and connecting teams around shared efforts, many featured tools were originally designed for very specific use cases rather than designed for cross-cutting adoption from day one.

For reuse to become a genuinely transformative strategy that accelerates AI adoption in government, it must be embedded into projects from the moment a pilot is conceived. European policy focus on interoperability is already steering public administration in this direction, and organisations like CTTI are leading the way by incorporating scalability criteria into solution design from inception. The mandate to reuse, and the requirements to make it possible, will become increasingly widespread across public institutions. The ultimate challenge, both for public administrations and the vendors selling to them, will be developing reusable solutions striking the right balance between standardised reuse and context-specific adaptation.

‍