Co-creation with civil servants is crucial for UTF.ai

Eight ambitious AI startups have been selected to compete in the AI Pitch Competition. The finals are on November 13, 2025, and IO+ will portray each contestant in the run-up to that event. The AI Pitch Competition is a Brabant-based contest that highlights the most innovative AI solutions, offering startups the opportunity to present their ideas, connect with industry leaders, and accelerate their growth. Today, we show what UTF.ai has to offer the world. Juul van Kessel (CEO & Co-Founder) and Giulio Tosato (CTO & Co-Founder) answer our questions about the ‘Schrijfkader’ tool.

What specific AI technology is at the core of your solution?

"The core of our innovation lies in a unique combination of practical AI and co-creation with civil servants. Schrijfkader ("Writing Framework") was not developed in a lab, but over a period of 1.5 years in collaboration with more than fifty civil servants from the municipality of Tilburg. As a result, the technology fits seamlessly into the daily reality of government organizations.

Technically, Schrijfkader is an intelligent Word add-in that provides automated feedback on structure, writing style, and regulatory compliance for each document type. Civil servants remain in complete control: the AI makes suggestions with explanations, but never makes decisions about content itself.

All writing guidelines are centrally managed on a platform, enabling changes to be implemented immediately and across the entire organization. In addition, Schrijfkader generates insight into common errors and improvement patterns, enabling organizations to continuously refine their guidelines and training programs."

How scalable is your AI solution?

"Schrijfkader is designed to be immediately scalable within existing government structures. Because it works as a Word add-in, it fits into familiar work processes without new software or complex implementation. Rollout can be easily scaled from 10 to 10,000 users within a single IT environment, with minimal support and a short onboarding video.

For each organization, we work with a licensing model based on population size, keeping costs predictable and manageable regardless of size. In addition, we assign a contact person or ‘ambassador’ for each department to bundle questions and encourage knowledge sharing, which accelerates adoption.

The biggest challenge lies not in the technology, but in public procurement procedures and the culture of change. That is why we deliberately start with small-scale pilots below the tendering thresholds, linked to measurable KPIs. This allows us to demonstrate the value, build support, and then scale up in a controlled manner to broad implementation within municipalities, provinces, and ministries."

How does your startup address potential ethical concerns related to bias, fairness, or transparency in AI decision-making?

"Our solution actually prevents ethical risks. It prevents civil servants from resorting to risky AI tools where data is never completely secure. With our solution, all data remains within the organization and nothing is disclosed externally. This ensures security and compliance with legislation. In addition, Schrijfkader enhances transparency: by checking and improving documents in advance, municipalities can publish more easily, creating an open government for residents.

It is important that people always remain in the loop. The add-in explains each suggestion and allows the user to accept or reject improvements. We check against explicit, locally established rules, and administrators receive usage signals so that they can improve rules and training where necessary. In this way, we combine security, transparency, and explainability in a single solution."

In what ways do you believe your AI solution can positively impact society?

"Schrijfkader directly contributes to better service, less work pressure, and more transparency in government. Civil servants save time by reducing the number of correction rounds and shortening the turnaround time for policy documents, resulting in faster and more consistent decision-making. This leads to higher-quality documents, better communication with the administration, and more room for service provision.

For residents, this means clearer letters, more understandable policies, and faster decisions. In addition, Schrijfkader makes it easier for governments to publish documents in accordance with the Open Government Act (Woo), which contributes to a more transparent and accessible government. For organizations, Schrijfkader speeds up new-employee training and enables the development of targeted training courses based on actual usage data: what is going well, and what is not?

We mitigate potential risks, such as overreliance on AI, by keeping human final control central. Users always receive an explanation for each suggestion and retain complete freedom of decision. To prevent rules from becoming outdated, organizations centrally manage their regulations with automated changelogs. We measure success not in AI activity, but in social value: faster decision-making, higher quality, and a government that communicates better with its citizens."

Tell us more about your entrepreneurial journey

"The biggest challenge in our startup phase was not the technology, but the adoption of AI within the government. Generative AI is new territory, and civil servants vary greatly in their attitudes: from cautious and risk-averse to enthusiastic and experimental. It was crucial to build trust and make it clear what AI does and does not do.

We solved that by focusing on transparency and co-creation. From the very beginning, we involved more than forty civil servants in the development, so that Schrijfkader was designed from the ground up for their daily practice. In each department, we appointed an ambassador who brought colleagues on board, collected feedback, and encouraged adoption.

Finally, we developed Schrijfkader so users don't notice that it uses AI. You don't need to prompt or have technical knowledge, making it accessible to everyone."

How are you preparing for the growing regulatory frameworks around AI, such as the GDPR, the AI Act, and other data privacy laws?

"Schrijfkader was developed from the initial design phase with privacy, security, and compliance as its starting point. The solution runs entirely within the government organization's existing IT environment, ensuring that all data remains within the secure government infrastructure and is not shared with external parties. As a result, Schrijfkader complies with the GDPR, GIBIT, and the upcoming European AI Act as standard, even if these regulations are further tightened in the future.

Humans are always in control: the Word add-in provides explanations for each suggestion, and an accept/reject system allows users to decide for themselves which improvements to adopt. The AI's functioning and limitations are also fully transparent thanks to clear help texts, disclaimers, and explainable models, ensuring responsible use.

Authorizations are directly linked to existing roles and rights within government domains, without the need for external SaaS services. This makes Schrijfkader a secure, compliant, and future-proof platform that enables innovation within the legal and ethical frameworks of the public sector."

Future vision: What is your long-term vision for your AI solution?

"Our current solution, Schrijfkader, proves how AI can support civil servants in making more impact and providing better services. By speeding up time-consuming writing and checking tasks, civil servants have more time for the work that really matters: improving policy, decision-making, and services to residents. Over the next five to ten years, we want to expand our platform with new AI solutions for the government, focusing on other repetitive and administrative processes. Always with the same principles: secure, explainable, and fully under human control.

Our ambition is to play a leading role in the transition to responsible AI in the public sector. We want to show that technology does not replace government but rather strengthens it by working more efficiently, transparently, and more people-orientedly. In short, we are building a future in which AI helps the government to make a greater social impact and continuously improve its services."

What opportunities does your location give you? What is still missing in the ecosystem you are part of?

"The Netherlands offers an exceptionally strong foundation for public innovation. The government actively invests in digitization and AI and is open to collaboration with young technology companies. This creates a climate in which initiatives such as UTF.ai can grow into concrete solutions that directly improve services.

Our location in Brabant further reinforces this: we are close to progressive municipalities and provinces, with short lines of communication to policymakers, knowledge institutions, and innovation programs. Thanks to this proximity, we can quickly set up pilots, collaborate intensively with civil servants, and test new applications directly in practice. In addition, we benefit from a strong national ecosystem, with networks such as the VNG, the Dutch AI Coalition, and Platform AI & Government, where knowledge sharing and collaboration are central.

What is still lacking is more sustainable structural funding for GovTech innovations. Many initiatives remain stuck in the pilot phase, even as there is a great need for scalable, secure AI solutions. The Netherlands has everything it needs to become a world leader in this field, provided we embed innovation structurally within the public sector."

The AI Pitch Competition: Why will you win this contest?

"We are UTF.ai, a team that not only talks about AI, but puts it to work in government practice. What sets us apart is not only our technology, but above all, the way we build: together with civil servants, for civil servants. In a year and a half, working shoulder to shoulder with the municipality of Tilburg, we have developed a solution that is already in daily use today. This is only possible if you really understand the work, the processes, and the people. We combine technical depth with administrative sensitivity, a rare combination in the AI world.

Our team consists of AI engineers, public administration experts, designers, and marketers who share one conviction: AI should strengthen the public sector, not replace it. We are young, multidisciplinary, and accustomed to turning complex issues into secure, explainable, and scalable solutions. Where others are still experimenting, we have proven that responsible AI can work within the government—securely, transparently, and with tangible results. From Brabant, we are building the next generation of GovTech innovations, and we are doing so with the same energy that made Schrijfkader great: down-to-earth, persistent, and impact-driven.

We are not winning this competition because we have the most beautiful slides, but because we are the strongest team to truly implement AI-driven solutions within the government of tomorrow. Secure, feasible, and with demonstrable social impact."