PresidencIA: Accelerating National AI Adoption

Why National AI Strategies Matter Now

We are at an inflection point. The World Economic Forum reports that 86% of employers anticipate AI will drive business transformation in the coming years [1]. Governments that fail to develop coherent AI strategies risk falling behind -- not just economically, but in their capacity to serve citizens, respond to crises, and shape the rules of a technology that will redefine how societies function.

Spain recognized this early. In 2020, the government adopted the Estrategia Nacional de Inteligencia Artificial (ENIA), one of Europe's first comprehensive national AI strategies. By May 2024, it was updated as the Estrategia de Inteligencia Artificial 2024, reflecting lessons learned and the rapid evolution of the technology landscape [2]. This was not a cosmetic refresh. The update incorporated four years of real-world implementation data, feedback from hundreds of stakeholders, and the emerging regulatory reality of the EU AI Act.

As a Senior AI Advisor and Program Director at the Office of the Prime Minister of Spain, I have had the privilege of working at the intersection of these forces -- translating national strategy into operational programs that bring AI capabilities into government at scale. The program we are building, which I refer to as PresidencIA, represents Spain's approach to making AI adoption in the public sector systematic, accountable, and genuinely transformative.

"The question is no longer whether governments should adopt AI, but how quickly and responsibly they can do so while maintaining public trust."

-- OECD AI Policy Observatory, 2024 [3]

The PresidencIA Vision: Three Pillars

PresidencIA is structured around three interdependent pillars that, together, create the conditions for sustainable AI adoption across the Spanish public administration. These are not abstract categories -- each maps directly to programs, budgets, and measurable outcomes.

Pillar 1: Community

Technology adoption in government fails when it is treated as a procurement exercise. The most common failure mode is deploying sophisticated tools into organizations where nobody knows how to use them, evaluate them, or govern them. PresidencIA addresses this head-on by building a community of practice among public officials.

We are mobilizing more than 500 public officials across ministries and agencies through structured training programs, cross-functional working groups, and a shared platform for experimentation. This is not about turning civil servants into data scientists. It is about building AI literacy at the level where decisions are made -- so that officials who commission, oversee, and regulate AI systems understand what they are working with.

Pillar 2: Applications

The applications pillar focuses on identifying, developing, and deploying AI solutions that address real operational needs within government. We prioritize use cases where AI can deliver measurable improvements in service delivery, operational efficiency, or crisis response capacity. Every application goes through a rigorous evaluation framework that accounts for technical feasibility, ethical implications, data availability, and institutional readiness.

Pillar 3: Infrastructure

You cannot build AI applications on sand. The infrastructure pillar ensures that the Spanish government has the compute capacity, data pipelines, and platform services necessary to support AI at scale. This includes investment in high-performance computing, shared data infrastructure, and secure cloud environments that meet the stringent requirements of government operations.

Spain's Strategic Position

To understand where PresidencIA fits, it helps to look at Spain's broader positioning in the global AI landscape. The OECD has identified six priorities within Spain's national AI strategy: scientific excellence, talent development, data and infrastructure, economic uptake, public administration adoption, and ethics and regulation [3]. This breadth is a strength -- it signals that Spain is not treating AI as a narrow technology initiative but as a systemic transformation.

In the Tortoise Media Global AI Index, Spain ranks approximately 21st globally overall, but notably achieves 4th place worldwide in government strategy [4]. That gap between overall ranking and strategy ranking tells an important story: Spain has built a strong policy foundation. The challenge -- and the reason PresidencIA exists -- is translating that policy strength into execution.

The Oxford Insights Government AI Readiness Index provides additional context. The United States leads globally at 87.03, followed by Singapore at 84.25. France, often cited as Spain's closest European peer, scores 79.36. The Western Europe average readiness score sits at 69.56 [5]. Spain is competitive within this band, but the data makes clear that readiness alone is not enough. The countries pulling ahead are the ones that convert readiness into deployed capability fastest.

Investment Allocation: Where the Money Goes

Spain committed EUR 500 million for the 2021--2025 period to fund its national AI strategy, with EUR 390 million already reaching beneficiaries as of the latest reporting [2]. This is a significant deployment rate and reflects the operational maturity of the strategy's implementation machinery, led by SEDIA (the Secretary of State for Digitalization and Artificial Intelligence).

Within this overall envelope, specific allocations reveal the government's priorities. EUR 50 million has been directed to high-performance computing (HPC), providing the foundational compute infrastructure that AI systems require. EUR 21 million is specifically earmarked for boosting AI uptake in the public sector -- this is the budget line that most directly supports PresidencIA's work. EUR 12 million targets AI talent development, while EUR 10 million supports the development of language models, a critical capability for a country that needs AI systems operating in Spanish and co-official languages [2].

From Policy to Practice: Mobilizing 500+ Public Officials

The single greatest lesson I have learned in this role is that government AI transformation is fundamentally a people problem. The technology is available. The budgets, while never unlimited, are allocated. What determines success or failure is whether the people inside government institutions develop the skills, confidence, and institutional support to actually use AI effectively.

This is why PresidencIA's community pillar is not an afterthought -- it is the foundation. We are building a network of more than 500 public officials who serve as AI champions within their respective organizations. These officials participate in structured learning programs that cover not just technical concepts, but also procurement, evaluation, governance, and the ethical dimensions of deploying AI in a public service context.

The approach is deliberately cross-functional. We bring together officials from health, transport, justice, taxation, and social services -- not because they share technical backgrounds, but because they share the challenge of applying AI within complex institutional environments. Some of the most productive insights have come from cross-pollination: a fraud detection approach developed for tax administration, adapted for social services; a natural language processing pipeline built for legal document analysis, repurposed for citizen inquiry routing.

"Spain's AI strategy encompasses six strategic priorities: boosting scientific excellence, promoting talent development, improving data ecosystems and infrastructure, driving economic uptake, enhancing public administration adoption, and establishing an ethical and regulatory framework."

-- OECD AI Policy Observatory, Spain Country Profile [3]

Agentic AI for Government: Beyond Chatbots

The most exciting -- and most demanding -- frontier in government AI is the move from narrow, prompt-response systems to agentic AI: systems that can reason, plan, use tools, and execute multi-step workflows with appropriate human oversight. This is where PresidencIA's applications pillar is focused.

Crisis Response

Consider crisis response. When a natural disaster strikes, government agencies must coordinate across dozens of organizations, process incoming data from satellites, sensors, social media, and field reports, allocate resources dynamically, and communicate with affected populations -- all under extreme time pressure. Traditional software helps with parts of this workflow, but the integration challenge is enormous.

Agentic AI systems can serve as coordination layers that monitor incoming data streams, flag anomalies, draft situation reports, propose resource allocation adjustments, and route communications -- while keeping humans in the decision loop for consequential actions. We are exploring these architectures within PresidencIA, with the understanding that government applications demand a higher standard of reliability, explainability, and auditability than commercial deployments.

End-to-End Product Lifecycle

Beyond crisis scenarios, we are applying agentic approaches to the full product lifecycle of government digital services. This means using AI not just to answer citizen queries, but to help design services, identify regulatory bottlenecks, draft policy proposals, model implementation scenarios, and monitor outcomes after deployment. The goal is to compress the time from policy intent to operational reality -- a cycle that in government can stretch from years to decades.

The EU Context: AI Act and Regulatory Sandbox

No discussion of AI strategy in Europe is complete without addressing the EU AI Act, which entered into force in August 2024 and is being phased in through 2027. Spain has been proactive in preparing for implementation. The Spanish AI Supervisory Agency (AESIA) became operational in June 2024, making Spain one of the first EU member states to establish a dedicated AI oversight body [6]. This was not symbolic -- AESIA is the institution responsible for enforcing the AI Act's requirements within Spain and representing the country in the European AI Board.

On March 11, 2025, Spain's national draft AI law was approved by the Council of Ministers, translating the EU framework into national legal instruments [7]. This law establishes the domestic regulatory architecture that complements the EU AI Act, clarifying competencies, procedures, and the role of AESIA as the national supervisory authority.

Perhaps most notably, Spain launched one of Europe's first AI regulatory sandboxes, selecting 12 AI projects for participation as of April 2025 [8]. The sandbox allows companies and public institutions to test AI systems in a controlled environment with regulatory guidance, generating practical insights that inform both the national and EU-level implementation of the AI Act. These 12 projects span health, finance, employment, and public services -- providing a diverse evidence base for regulatory learning.

"Regulatory sandboxes are essential instruments for translating the AI Act's risk-based framework into practical guidance that organizations can actually implement."

-- Adapted from EU AI Act, Article 57: AI Regulatory Sandboxes [9]

Looking Forward

The next phase of PresidencIA focuses on three imperatives. First, scaling the community beyond the initial 500 officials to create a self-sustaining ecosystem where AI literacy propagates organically across the public administration. This means training trainers, building internal case study libraries, and establishing peer review mechanisms for AI projects.

Second, hardening the applications pipeline. The initial wave of AI projects has generated valuable lessons about what works in government contexts. We now need to codify those lessons into repeatable frameworks -- standardized evaluation criteria, deployment checklists, monitoring dashboards, and incident response protocols specifically designed for public sector AI.

Third, deepening the infrastructure layer. Spain's EUR 50 million investment in HPC is a strong start, but the compute demands of modern AI -- particularly large language models and agentic systems -- are growing faster than anyone predicted. We need to ensure that government teams have access to the compute resources they need without creating dependency on any single commercial provider.

There is also a broader strategic dimension. As the EU AI Act enters full enforcement, Spain's early investments in AESIA, the regulatory sandbox, and the national AI law position the country to be a rule-shaper rather than a rule-taker within Europe. The 12 sandbox projects are not just compliance exercises -- they are generating the empirical evidence that will inform how the AI Act is interpreted and applied across the entire European Union.

The work is far from finished. AI adoption in government is a marathon, not a sprint. But I am convinced that the approach we have built with PresidencIA -- grounding strategy in community, prioritizing real applications over pilot theater, and investing in infrastructure that outlasts any single technology cycle -- gives Spain a solid foundation. The goal is not to be the country that talks most about AI in government. It is to be the country that actually delivers it.