EDIH CTU wins the AI Awards 2025 in the Public Administration category

0
9111

The European Digital Innovation Hub at ČVUT succeeded in the prestigious AI Awards 2025 competition, held annually by the Česká národní AI platforma (CNAIP). Out of 120 nominations across six categories, the expert jury selected projects that best illustrate how artificial intelligence is becoming a practical tool shaping education, research, public administration, and business. The jury highlighted the internal knowledge chatbot ADAM, developed by EDIH CTU together with experts from the Český institut informatiky, robotiky a kybernetiky CIIRC ČVUT for the Auditní orgán Ministerstva financí to speed up work with extensive documentation related to EU funds.

EDIH CTU received votes from an independent jury composed of Jindřich Fryč, the highest-ranking state official at the Ministry of the Interior, Jan Kavalírek, Deputy Minister of Industry and Trade and Government Envoy for Artificial Intelligence, and Eva Pavlíková, Director General of the IT Section at the Ministry of Regional Development. The award confirms that connecting top-tier research produced at ČVUT with the real needs of public administration brings results visible not only in the Czech Republic but also across Europe.

Chatbot ADAM: a verifiable assistant in an overwhelming number of regulations

In cooperation with researchers from CIIRC ČVUT, the Alquist Insight application was created to make it possible to quickly and verifiably search large document collections. Based on this technology, the ADAM tool (Automatizovaný Digitální Asistent pro Monitoring) was developed for the Auditní orgán Ministerstva financí. It helps auditors navigate more than 40,000 pages of regulations related to European programmes. ADAM significantly speeds up audit processes, unifies interpretation of methodologies across teams, and contributes to more efficient public administration as well as greater transparency and accuracy in the use of EU funds.

“The project was primarily designed for the Auditní orgán Ministerstva financí, but it can also be used by the other participating bodies. We expect it to make it easier to navigate all the documents of the various ministries involved,” says Iva Hrůzová from the Auditní orgán Ministerstva financí, who coordinated development and testing. The user perspective was overseen by Hana Horejsková, head of the Department of Coordination of ESIF and Other Funds Audits. The startup Promethist.ai also contributed to the development by ensuring robust implementation and solution stability.

Security and verifiability were essential. “We designed ADAM as a secure and verifiable assistant. Answers always refer to specific documents so auditors can immediately see the source of each conclusion,” explains Jan Šedivý, head of the expert team at CIIRC ČVUT. The Big Data & Cloud Computing team—Tomáš Brich, Deniz Gökdeniz, Stanislav Khvedynich, Matěj Klouček, and Max Tyx—designed the RAG (Retrieval Augmented Generation) architecture, knowledge-base management, and accuracy-testing methods. Thanks to the joint work of EDIH CTU, CIIRC ČVUT, and Promethist.ai, the solution is ready for further secure deployment in public administration.

From cities to hospitals: EDIH CTU’s broader portfolio of public-sector projects

Beyond the ADAM project, EDIH CTU is involved in dozens of other solutions that help modernize Czech public administration. For the City of Prague, it built a smart waste-collection system where container-fill prediction reduces unnecessary trips. For the Policejní prezidium České republiky, EDIH CTU is developing a tool to monitor energy consumption using a sensor network that detects waste in real time. At the Fakultní nemocnice v Motole, the DigiDoc and NuklKalibr services automate administration and strengthen patient safety. The Státní zdravotní ústav uses a tool for evaluating ergonomic working postures based on 3D reconstruction.

Other projects target social care, transportation, and education. The Gerontologické centrum uses haptic-cognitive AI tasks for senior training; Správa železnic is testing recognition of rail-infrastructure elements from train-mounted cameras; Základní škola Zlíchov is piloting a specialized chatbot for neurodivergent pupils; and the Západočeská univerzita uses a service that links students with experts. Research institutions such as the Matematicko-fyzikální fakulta Univerzity Karlovy and the Ústav jaderné fyziky Akademie věd České republiky use AI tools for automated analysis of experimental data, new-materials design, and faster, more accurate task processing.

“Our goal is to create pilot solutions and then support the transfer of technologies from pilots into real operation. ADAM and other projects show that when clear requirements, the Test Before Invest method, and responsible design come together, the result is services that genuinely help authorities and citizens. Our experts can scale them to other agendas, not only at the Ministry financí,” says Barbora Zochová, Director of EDIH CTU.

EDIH CTU also supports the long-term adoption of AI in state and public administration through conferences, meetups, training, and pilot projects. These include the traditional AI4GOV event, organised together with prg.ai as part of Dny AI, which shares concrete practices from data work to citizen-facing service impact.

The award for EDIH CTU wasn’t the only recognition received by teams connected to CIIRC ČVUT in this year’s AI Awards. Among the finalists in the “AI in Research and Development” category was also the DreaMS project by brothers Roman and Anton Bushuiev from the Ústav organické chemie a biochemie AV ČR and CIIRC ČVUT. The DreaMS machine-learning model accelerates the analysis of previously unknown molecules by interpreting mass spectra. The project, published in Nature Biotechnology, was developed in collaboration with Josef Šivic from CIIRC ČVUT, who also served on the jury of this year’s AI Awards.

Previous articleOlga Štěpánková: I feel responsible for AI
Next articleCLARA: Artificial Intelligence and quantum computing in systems neuroscience research