Successfully deploying AI in the Dutch public sector
The successful deployment of AI in the public sector lags behind that of other industries, according to EY. AI is evolving faster than government agencies can adapt, and there are challenges related to governance and transparency. On top of this, the Dutch government does not know if many of its artificial intelligence (AI) systems work as intended. As a result, government agencies struggle to move beyond experimenting with AI, which causes a pilot-to-scale gap.
What are the main challenges causing this gap, and how can government institutions bridge it? Let’s take a look.
Challenge 1: Legacy IT
In 2025, the Dutch government presented the Netherlands' Digitalisation Strategy. It states, among other things: "The government is currently struggling with problematic legacy issues, services that are still insufficiently aligned with the needs of citizens and entrepreneurs, fragmented policies, and the need to strengthen digital security and resilience."
The digital situation in the Dutch public sector is the result of years in which each institution and organisation was able to make its own digital choices without a clear, top-down policy.
Add to this the fact that the government is the largest consumer of digital services, and you can understand the enormous impact of this situation. Every resident of the Netherlands interacts with the government, making poorly functioning digital processes more noticeable than in commercial organisations. This means AI pilots must perform perfectly if they’re to achieve large-scale deployment.
Solution: A single AI platform
Using a unified AI platform can create a government-wide infrastructure. Decoupling AI applications from outdated core systems through APIs and middleware helps prevent pilot projects from getting bogged down by legacy issues. This mitigates the need for repeated technical and legal reinventions, allowing for faster scaling within and between government organisations.
Challenge 2: Unclear AI policy
The lack of clarity surrounding AI policy is also affecting the public sector. There’s considerable debate about digital sovereignty, control over data, and transparency.
This leads to a situation where not all AI capabilities can be used, meaning citizens don't get the same smooth experiences they're used to with non-governmental organisations. Furthermore, it increases the risk of shadow AI—solutions used by government employees outside the IT department's purview to perform their work more efficiently.
The Netherlands’ Digitalisation Strategy also recognises this problem and points to the need for a comprehensive policy. However, the precise form of that policy remains unclear.
Solution: A clear framework
Ensure an explicit AI policy that provides a clear framework for each risk class. Use a list of permitted solutions and applications that teams can experiment with. Combine this with transparency requirements (logging, explainability, audits) that enable scaling rather than blocking. This helps decrease shadow AI and create more room to scale up successful pilots.
Challenge 3: Lack of AI skills
Research by Binnenlands Bestuur reveals that more than 75% of decision-makers in the public sector lack full confidence in the digitalisation drive within their organisation.1 The ServiceNow Enterprise AI Maturity Index shows that globally the public sector is behind other industries in terms of the percentage of frontrunners in AI. Technology (32%) and manufacturing (21%) are well ahead of the public sector (15%).
The Dutch public sector’s struggle with AI seems to be part of a bigger development. The European Commission's State of the Digital Decade 2025 report outlines the situation for the Netherlands: "Most Dutch organisations have a basic level of digitalisation, but smaller organisations in particular are struggling to keep up with the pace of adoption of new technologies—especially when it comes to AI."
Solution: Multidisciplinary approach
Approaching AI from a multidisciplinary perspective and investing in employees’ AI skills can create a work environment in which AI becomes part of regular work processes. Successfully scaling AI projects should be the responsibility of all departments, not just IT. Public sector organisations can learn from initiatives taken by organisations in other sectors to better integrate AI into daily operations.
Mature AI deployment
The pilot-to-scale gap is not a purely technical problem. It’s the result of deeper structural, organisational, and policy choices.
The public sector is already taking significant steps towards mature deployment of AI. Most notable are various government initiatives using on-premises AI, such as SSC-ICT with Vlam2 and the Municipality of Amsterdam with ChatAmsterdam.3
These examples demonstrate that AI doesn't necessarily have to originate in the cloud. It can also run in a self-managed, physical environment. The ServiceNow AI Platform can support both cloud and on-premises AI applications.
Find out how ServiceNow helps put AI to work for the public sector.
1 Binnenlands Bestuur, Digital transformation in the public sector not yet as far as hoped, Sept. 22, 2025
2 SSC-ICT, Vlam.ai is spreading like wildfire, March 2025
3 Municipality Amsterdam, Amsterdam launches ChatAmsterdam: Its own AI assistant for civil servants, May 19, 2025