Monitoring and auditing the targeted use of budget funds, including loans, is an important task that is performed manually in many municipal, regional, and even federal agencies.

Monitoring contractors, the quality of their work, and ensuring that the bid price matches the actual costs incurred are activities our client is required to perform regularly. Meanwhile, the budgetary organization is forced to bear responsibility for the misuse of funds. This wastes time, resources, and the potential risk of significant fines. Not to mention potential legal costs and the search for alternative contractors.

An automated solution capable of offloading most of the audits would help the client protect itself and save budget funds by avoiding hours of manual audits.

We have developed an automated solution based on AI agents that not only significantly speeds up audits but also automatically generates necessary materials for presentation to the prosecutor’s office and investigative bodies, as well as press releases for the media if the audit reveals violations.

The solution is based on a system of AI agents running on a secure cloud infrastructure and a frontend implementing an NLP interface for working with uploaded reports and documents using natural language commands.

Testing the MVP on historical cases and commission findings showed that the AI’s conclusions regarding suspected misuse of funds match human conclusions 99% of the time.

The implementation of the pilot solution has successfully automated and accelerated the routine work of reading reports, reconciling figures, and preparing submissions, eliminating human error. The client has successfully saved dozens of hours of work, and the accuracy of decisions has increased. Plans call for further scaling of the solution.