A European Transition in Progress
For more than three decades, governments have pursued digital transformation.
The first phase focused on digitization – moving paper-based procedures online, still in progress …
The second phase emphasized interoperability, shared platforms, and cross-agency integration.
We are now entering a third phase, deeper and more consequential: the Agentic State.
This transition is not about better interfaces. It is about delegation.
What Do We Mean by Agentic State ?
An Agentic State is one in which software agents are authorized to act on behalf of public institutions, within clearly defined legal and political mandates.
These agents can:
execute administrative decisions,
trigger legally binding actions,
coordinate workflows across agencies,
support compliance, crisis response, and procurement at scale.
Humans remain central – but their role shifts toward oversight, accountability, exception handling, and policy judgment.
This is an architectural shift. It requires:
law that can be executed,
interoperable registries,
strong digital identity,
governance mechanisms for accountability, safety, and redress.
Why This Shift Is Happening Now
Two forces are converging.
First, state complexity has reached a level where human-only administration no longer scales – especially in welfare, taxation, compliance, crisis management, and public spending.
Second, AI systems are no longer confined to pilots or experimentation. They are now capable of being embedded inside core government workflows.
The key question is no longer whether AI can be used in government, but where delegation is legitimate and how it is governed.
This is where Europe shows clear internal differences.
Nordic and Mediterranean Europe: Two Administrative Trajectories
Nordic countries – Denmark, Sweden, Finland, Norway, and often Estonia as a close reference – moved earlier toward agentic models because their institutional foundations were already aligned:
coherent administrative structures,
high levels of institutional trust,
and a tradition of translating public law into executable systems.
In these contexts, agency is often invisible. The state simply works.
Mediterranean countries followed a different path. Digitalization has been significant, but delegation remains limited. AI is often positioned as support rather than authority, due to:
institutional fragmentation,
layered legal interpretation,
and a higher sensitivity to automated decision-making.
This is not a question of ambition, but of governance DNA.
A thoughtful roundtable …
It is in this context that I was invited, a few days ago, to participate in this conference on The Architecture for the Agentic State.
The composition of the panel is itself meaningful.
Participants include:
representatives from the Government Offices of Sweden, where agentic approaches are already embedded in public administration;
officials from the European Commission, directly involved in shaping the regulatory and governance framework for AI in the public sector;
voices from the World Bank, bringing a global, capacity-building perspective on state transformation;
leaders from cities such as Amsterdam, Warsaw, and Kongsvinger, where agentic principles are being tested at municipal level;
and academic and policy institutions like the Hertie School and Zorginstituut Nederland, which work at the intersection of governance, law, and technology.
This was not a discussion about tools.
It was a conversation about state capacity, institutional architecture, and accountability in an age of delegation.
My professional journey has unfolded across regulated industries, large-scale infrastructures, and public-sector transformation – long before the term Agentic State existed.
I began working with systems where software was already entrusted with critical responsibilities:
identity, billing, access control, network operations, and later public administration.
Across decades of work in telecom, payments, and government, I have consistently observed at the boundary between:
human judgment,
and machine-executed processes under formal rules.
Nordic contexts, in particular, shaped my understanding of what is possible when:
systems are allowed to function end-to-end,
automation is mandated rather than symbolic,
and accountability is designed into the architecture.
At the same time, my work in Southern Europe has kept me grounded in the realities of governance systems where delegation must be earned, negotiated, and justified step by step.
A Personal Perspective
I see my presence in this forum not as representation of a single model, but as a connecting role.
Between:
Nordic systems that already operate agentically,
and Mediterranean systems that are still defining the conditions under which delegation becomes legitimate.
Between:
law as executable code,
and law as institutional interpretation.
A Closing Thought
The Agentic State is not a technological destination.
It is an institutional transition.
Nordic countries show what is possible when trust, coherence, and delegation align.
Mediterranean countries remind us that legitimacy, culture, and governance cannot be bypassed.
This is why this discussion – and this roundtable – matters now.
References:
AgenticState.org – The Core Initiative
The official site describing the platform that supports governments in moving toward agentic AI-enabled public services. The Agentic State
Their Vision Paper outlines the architecture and enablement conditions for an Agentic State, covering governance, data, cybersecurity, culture, procurement, and more. The Agentic State
Tallinn Digital Summit 2025 Launch
Announcement of “The Agentic State: Rethinking Government for the Era of Agentic AI” — a foundational paper presented with contributions from digital government leaders globally, connected to the World Bank and the Global Government Technology Centre Berlin. The Agentic State
Global Government Technology Centre & Whitepaper
A related whitepaper introduction from the Global Government Technology Centre discussing how agentic AI will transform core public administration functions. globalgovtechcentre.org
Related Academic & Governance Discussion
A research paper on oversight structures for agentic AI in public-sector organizations, which discusses governance challenges that intersect with the Agentic State idea. arXiv
