Is the Netherlands ready for the digital civil servant?

That government is not science fiction. In Abu Dhabi, virtual civil servants already handle services proactively. In Singapore, much interaction with citizens is managed by AI, with satisfaction rates as high as 98%. The technology exists, is mature and operates at scale. The real question is not whether the technology is ready, but whether the Dutch government is.

The public debate about AI in government remains focused on generative applications: drafting policy papers, summarisingmeetings and helping with emails. Useful, efficient – but ultimately just a faster typewriter. The process remains unchanged, is just less time-consuming.

The real step forward comes with agentic AI. This is no longer about chatbots generating text, but digital colleagues who can independently execute tasks, orchestrate processes across systems and organisations, prepare decisions and – within defined boundaries – make them.

This is precisely the level of capability needed in a labour market expected to face a shortage of around one million workers by 2035. The private sector is already preparing: banks and multinationals are redesigning operating models, job descriptions and cost structures. Many are moving towards “zero-ops”, shifting routine work to AI agents so scarce talent can focus on customer interaction, exceptions and value creation. The public sector must follow suit before labour shortages and unsustainable cost structures make the system unworkable.

Why is it not happening yet?

There is no shortage of pilots, sandboxes and exploratory projects within government. AI assistants are being tested, research partners are involved, HR teams are exploring implications and national strategies are in place. According to the AI Radar 2026, most employees already use AI, with nearly a third doing so daily.

Yet progress remains limited. Not because the technology performs worse in the Netherlands than elsewhere, but because the governance and technological foundations are not ready. More than 70% of employees have not received AI training.

A risk-averse governance model has emerged, where every risk must be mitigated upfront. As a result, AI is primarily viewed through the lens of what might go wrong, rather than what demonstrably improves when its risks are managed intelligently.

At the same time, government still relies heavily on legacy IT. Data remains siloed, integrations are fragile and real-time access is limited. Agentic AI depends on clean, connected and accessible data. Without modernisation, AI will remain an experiment.

Finally, organisations often respond to challenges by adding more people. That approach may work in the short term, but is unsustainable in a structurally tight labour market. The real question is not how to fill vacancies, but how to redesign work using mixed teams of humans and AI agents.

Agentic AI and the human touch

There is a common fear that AI will make government less human. Evidence suggests the opposite. In Singapore, AI-supported service centres report higher satisfaction among both citizens and employees. Routine tasks are handled by machines, allowing people to focus on providing empathy and support.

Imagine a social worker spending time with families instead of completing paperwork. Or a caseworker starting the day with a clear overview of risks, missing information and suggested actions. Or a citizen no longer battling bureaucracy because the system proactively removes barriers and prevents errors.

This is how agentic AI can strengthen the human dimension.

Waiting means falling behind

The transition to an AI-enabled government will not happen in one leap, but through deliberate design choices. Leaders must distinguish between low-risk and high-risk use cases, invest in data as infrastructure and treat AI agents as digital colleagues with clear roles, lines of accountability and oversight.

Agentic AI is coming regardless. Employees already use it – often outside official channels. The real choice is whether to shape how humans and machines work together, or leave it to fragmented initiatives and growing labour shortages.

A proactive, efficient and empathetic government is not a distant vision. It is a design choice.

Read the article in Dutch in Platform O.