Name in original language
Digitalni asistenti u javnoj upravi
Initiative overview
The initiative aims to strengthen the accessibility and efficiency of public services in Serbia through the deployment of sovereign AI assistants developed and operated by the Serbian Office for IT and eGovernment. The rollout began in 2025 with the introduction of an AI-based digital assistant on the national eGovernment Portal, enabling citizens to obtain information and guidance in clear, everyday language.
Since its initial deployment, the initiative has expanded to support information provision for over 50 digital public services. Apart from the national eGovernment Portal, assistants have been implemented for the Ministry for Family Welfare, the property rights registration process, and the Foreigners Portal, where they provide support on topics such as visa applications and residence procedures. These deployments have demonstrated the system’s capacity to respond to high volumes of citizen enquiries across multiple service domains.
The initiative is expected to evolve from primarily informational AI assistants towards transactional capabilities that allow users to complete administrative actions digitally. This next phase aims to further simplify procedures and improve digital accessibility. Supported by Serbia’s sovereign large language model development and ongoing infrastructure upgrades, the initiative is designed for institutionalisation across public bodies and scalable adoption throughout the public sector.
Results, outcomes and impacts
The initiative has delivered measurable improvements in service accessibility and administrative efficiency. Within two weeks of deployment, enquiries to the Contact Centre decreased by 20%. AI assistants now provide information for around 50 e-services and support high-demand domains, including family welfare, property registration and migration processes, with nearly 200,000 recorded conversations.
These results demonstrate strong adoption and reduced manual workload. Future impacts are expected through transactional AI capabilities that will further streamline procedures and enhance service delivery.
Challenges and lessons learned
Early risks related to accuracy, user trust, and safeguarding personal data were mitigated through phased deployment, continuous model evaluation, and strict adherence to security and privacy standards. A major lesson learned is the importance of clear governance, strong coordination across public bodies, and transparent communication with users.
Success requires reliable national infrastructure, access to high-quality local datasets, and sustained institutional commitment to responsible and ethical AI development.
