Platform for Monitoring State Financial Management Using AI

The project addresses the limited capacity of civil society, media, and the wider public to systematically monitor how public money is spent. In Slovakia, data on public contracts, procurement, subsidies, companies, beneficial owners, and public institutions are available in several public registers, but they are fragmented, technically difficult to process, and often lack mutual context. As a result, journalists, analysts, and watchdog organisations must manually search, clean, and connect data from different sources, which makes public oversight slow, selective, and dependent on limited human capacity.
This problem is especially relevant in the area of public spending, where risks such as non-transparent contracts, concentration of suppliers, conflicts of interest, weak competition, or links between private companies and public institutions may remain unnoticed. Although open data formally exist, their practical usability for democratic oversight is limited. The gap between the amount of available public information and the ability of civil society to use it creates an information asymmetry between public authorities and citizens.
The challenge is also technological. Existing watchdog and investigative work is not yet sufficiently supported by modern data infrastructure and AI-based tools that could help identify patterns, red flags, and connections across large volumes of documents. Without such tools, many potential issues in public spending cannot be detected at scale.
WatchGov responds to this challenge by developing a modular data platform that will connect key public datasets and make them easier to search, analyse, and use in investigative and watchdog work. The project will help media, civil society organisations, and the public to monitor public finances more effectively and to strengthen transparency, accountability, and democratic control.

The project aims to develop a pilot version of WatchGov, an AI-supported modular platform for public oversight of state finances. The platform will connect and structure key public datasets so that journalists, watchdog organisations, analysts, and the public can more easily search, analyse, and identify risks in public spending.
The first main activity will be the design of the data architecture and methodology for collecting and processing public data. This will define how contracts, subsidies, procurement data, companies, beneficial owners, debts to the state, and related documents will be structured and linked.
The second activity will focus on collecting, cleaning, and processing data. The project will extract and process approximately 700,000 contracts from the Central Register of Contracts and 50,000 public procurement documents, using OCR and LLM-based extraction where needed. This will create the core database of contracts, subsidies, and public financial flows.
The third activity will connect additional data modules to the database, including public procurement data, company financial results, beneficial ownership data, and debts to the state. This will make it possible to identify links between public institutions, suppliers, owners, and financial risks.
The fourth activity will be the web design and development of the user interface. The pilot platform will include advanced search, filters, visualisations, entity links, red flags, and an API for further analytical use.
The project will also include ongoing consultations with representatives of at least five media or watchdog organisations, who will help test the platform and define practical investigative scenarios.
Finally, the project will produce at least three investigative and analytical outputs based on WatchGov data and AI agents. These outputs will demonstrate the practical use of the platform and will be published through WatchGov and, where possible, partner media.

The project aims to achieve three main results. First, it will create a functional pilot version of the WatchGov platform, providing a structured and searchable database of public financial flows. This will include processed data from public contracts, procurement documents, and selected public registers, connected through a modular data architecture. The result will be a practical tool that makes fragmented public data easier to access, understand, and use.
Second, the project will strengthen the capacity of journalists, watchdog organisations, and analysts to monitor public spending. Representatives of at least five media or watchdog organisations will be involved in consultations, testing, and the definition of investigative scenarios. They will benefit from a tool designed according to their practical needs, including search functions, filters, visualisations, red flags, and an API for further analytical work.
Third, the project will demonstrate the practical use of AI-assisted watchdog work by producing at least three investigative and analytical outputs based on WatchGov data and AI agents. These outputs will help identify risks, patterns, or links in public spending and will be presented in a form understandable to the wider public.
The main beneficiaries will be civil society organisations involved in watchdog and advocacy activities, investigative and data journalists, analysts, and experts working with public data. The wider public will also benefit from better access to information about how public money is spent and from published analytical outputs. In the longer term, public institutions may benefit as well, as the project can provide evidence and feedback useful for improving transparency, public procurement, open data policies, and anti-corruption measures.