Ministry of Digital Transformation of Ukraine delegation receiving the Best Cases Award in Copenhagen (DK) on 26 November 2025
The Responsible Organisations
The AI Centre of Excellence at the Ministry of Digital Transformation of Ukraine is a publicly founded hub for the development of artificial intelligence and its integration in public administrations, as well as in strategic private sectors. It is the organisation deputed to coordinate the development and launch of Diia.AI.
The Ministry of Digital Transformation of Ukraine was awarded with the Government-to-Citizens Best Cases Award at the 2025 SEMIC Conference.
Diia is an inhouse government technology company belonging to the Ministry of Digital Transformation of Ukraine. Its role is to provide technical, organisational and operational support to the digital-services infrastructure under Diia, including Diia.AI.
1. The problem
Modern public administrations increasingly face the challenge of using emerging technologies to address long-standing bureaucratic inefficiencies. In Ukraine, this need was particularly pressing, as citizens were often required to navigate complex websites, switch between multiple portals, and complete lengthy forms just to access essential public services. These procedures were slow, inconvenient, and still relied heavily on in-person interactions, which not only prolonged service delivery but also created opportunities for improper practices, including corruption. Consequently, the user interaction with state authorities remained neither intuitive nor easily accessible.
2. The solution: an AI agent for government services
The Ukrainian Ministry of Digital Transformation, together with Diia, an Ukrainian inhouse government technology company, and some private tech industry partners (i.e., KitSoft and Google) developed Diia.AI, an artificial intelligence agent that, beyond answering questions, provides citizens with governmental services directly within a chat, in order to simplify how they engage with public administrations by turning complex bureaucratic procedures into an accessible, immediate, user-friendly conversation.
The service enables individuals to receive information and access public services through a natural, conversational interface on their digital devices. Moreover, the implemented use case aligns with Ukraine’s broader efforts to build a user-centric digital government. By reducing the need for in-person contact, it also contributes to more transparent service delivery and helps limit risks of corruption. Overall, the service enhances accessibility, efficiency, and trust, which are important elements in the evolution of a modern digital state.
2.1 Technology
The technological innovation achieved by Diia.AI consists in an advanced agentic architecture, that moves beyond the traditional question-answering chatbots and enables the system to execute multi-step actions on behalf of the user. Furthermore, its design is rooted in a hybrid cloud model that balances security, performance and scalability, an approach particularly suited to public-sector digital services.
Among the key innovations brought by the AI agent from the Ukrainian government, the following are included:
-
The use of LangGraph, an orchestration framework for building and managing AI agents, enables Diia.AI to handle complex user interactions, representing a significant evolution compared with conventional linear LLM pipelines.
-
The reliance of the system on a hybrid infrastructure that ensures scalability thanks to Google Vertex AI (Gemini 2.0 Flash) and allows government systems to retain full control over sensitive information while benefiting from cutting-edge AI capabilities.
-
The use of a four-stage Personal Identifiable Information (PII) protection mechanism. Personal data is anonymised and transformed into placeholders before any interaction with the LLM, ensuring that the model never processes individuals’ identifiers. A secure backend service reinserts the correct data after the AI generates a response, with an innovative approach for privacy-preserving AI in the public sector.
-
The incorporation of custom multilayer guardrails, combining keyword filtering, machine-learning classification, and LLM-based reasoning. These guardrails are optimised for the Ukrainian language and protect the system from prompt-injection attempts and harmful content generation.
2.2. Integration within the Diia.Engine ecosystem
Ukraine’s artificial intelligence agent is incorporated in Diia.Engine, a national, open-source, low-code platform that utilises emerging technologies to power secure, interoperable, and citizen-centric public services. The platform has been developed by Diia starting from 2023, during a period of national crisis, with the goal of transforming Ukraine’s fragmented, paper-based administrative landscape into a modern, scalable, and transparent digital ecosystem.
Thanks to Diia.Engine, citizens have been able to connect with national institutions at all levels, from central governmental entities to local authorities, in order to access over 150 public services, including digital identification through eDocument and relocation assistance during the Russian attack with eRecovery. As a matter of fact, a digital environment like the one provided through the platform proved essential for delivering public services in a country facing full-scale invasion, missile strikes, curfews, and mass displacement, all of which made traditional paper-based processes unworkable. Thus, crises did not slow digitalisation, they made accelerating it even more necessary.
Within this framework, Diia.AI was created to streamline access to public services offered on the national platform with the support of an AI assistant, using a highly agile and user-centric implementation process. As a matter of fact, after analysing the twenty most requested government services available through Diia.Engine, those with the greatest potential impact in terms of ticket volume were selected. The entire user flow was then optimised according to conversational and intuitive design principles, and pilot projects were developed for rapid iteration and feedback. Diia.AI was subsequently launched in an open beta phase, actively collecting user feedback to continuously train and improve the assistant’s performance based on real-world interactions.
Besides strategic technology partners, the whole project involved deep collaboration with the GovTech ecosystem, demonstrating Ukraine’s evolving approach to public sector innovation, as it is described in Iuliia Drobysh’s (Global Government Technology Centre Kyiv) opinion piece published for the Public Sector Tech Watch collection.
2.3. Interoperability, transparency and scalability
Diia.AI has been developed with interoperability principles as foundational requirements, following the European Interoperability Framework wherever applicable. Among other specificities, technical interoperability with government registries is secured through the Model Context Protocol (MCP), an open-source tool that connects AI applications to Ukraine’s data infrastructure. At the same time, semantic interoperability for all information ingested into the RAG knowledge base is ensured by the use of a uniform JSON metadata format, while organisational interoperability is the result of a collaboration among the responsible organisations involved.
The AI agent strengthens government transparency by giving citizens continuous access to official information and clear explanations of administrative procedures. In fact, its citation system links each AI generated output to verified source documents, and users can therefore validate responses independently. While the core system cannot be open sourced due to its integration with protected state registries, the project team remains committed to sharing architectural guidance, security principles and best practices, supported by the wide use of open-source technologies in the underlying stack.
Scalability and reusability were central design goals of Diia.AI. As a matter of fact, the system architecture allows each component to scale independently, while LLM inference through Vertex AI supports high performance for large scale usage across the country. Although reuse has not yet begun, the architectural blueprint, including the standardised implemented MCP protocol and the agentic AI framework, is intended for broader adoption across Ukrainian public administration and potentially by other governments.
3. Key Benefits
Dii.AI, by integrating a chatbot that delivers government services within the Ukrainian national public administration Diia platform, can offer several combined benefits:
-
Enhance citizen experience by turning complex bureaucratic processes into intuitive conversational interactions. Shifting from an approach in which users are obliged to navigate complex websites and be involved in time-consuming operations like filling forms to one that requires only chatting with an AI chatbot, has already proven to save time and furtherly foster digitalisation. As a matter of fact, early adoption metrics show strong user trust and satisfaction (CSAT score of over 80%), rapid uptake, and substantial time savings, with the potential to reduce service resolution times to under three minutes and save more than one million citizen-hours annually.
-
Strengthen government efficiency and transparency. Developing AI-facilitated operations through a chatbot agent allows to automate high-volume requests, reduces support burdens, and provides verifiable, source-linked information. As a matter of fact, it is projected a 30% reduction in human support for tickets. Moreover, Diia.AI’s pioneering privacy design and secure hybrid architecture also establish a new standard for trustworthy AI in the public sector.
4. Future challenges
As it was underlined by the AI Centre of Excellence at the Ministry of Digital Transformation of Ukraine, the increasing adoption of Diia.AI for administrative services will require addressing some challenges:
-
Ensure scaling and operational resilience. As adoption grows nationwide, maintaining consistent performance and reliability will require sustained investment in scaling, monitoring, and optimisation. The hybrid dependency on external LLM infrastructure may also pose long-term cost, capacity, and sovereignty considerations.
-
Adapt to interoperability and openness requirements. Ensuring reusability and further alignment with European regulations will require substantial additional work on documentation, interoperability compliance, and governance.
Website and Contact Information
Useful links:
-
Diia.AI explained by the AI CoE of the Ministry of Digital Transformation of Ukraine: https://digitalstate.gov.ua/news/govtech/diiaai-pershyy-u-sviti-derzavnyy-ai-ahent-iakyy-ne-prosto-konsultuye-a-nadaye-posluhy-iak-pratsiuye-shtuchnyy-intelekt-na-portali
-
Diia.Engine on the Interoperable Europe Portal (DG DIGIT): https://interoperable-europe.ec.europa.eu/collection/egovernment/solution/diiaengine-open-source-low-code-platform
-
PSTW Opinion Piece by Iuliia Drobysh: https://interoperable-europe.ec.europa.eu/collection/public-sector-tech-watch/news/iuliia-drobysh-pstw
Project contact:
Detailed Information
Case Viewer ID: PSTW-2639
Year: 2025
Status: Implemented
Responsible Organisation: Ministry of Digital Transformation of Ukraine
Geographical extent: National
Country: Ukraine
Function of government: General Public Services
Technology: Artificial Intelligence
Interaction: G2C