In legacy financial management systems, high-volume transaction matching is often a complicated and time-consuming task despite automation involved. The following example will help understand this complexity, the required human efforts to resolve financial and accounting errors on a large scale and the need for a solution that saves time, cost and improves efficiency.

At the Federal Government’s largest agency, the Department of Defense (DOD), it costs millions of labor hours per year to resolve erroneous financial transactions. This is really surprising but this is true. One of the DOD’s entities, the Army alone experienced somewhere between two and three million of unmatched transactions (UMTs) in the 2019 fiscal year while on average the Army Financial Management and Control (FM&C) team identifies and resolves over one million UMTs per year.

Enters the AI

The challenge before the DOD was that the existing robotic process automation (RPA) infrastructure was found to be limited in its capabilities. In order to resolve this pertinent issue of accounting discrepancies in the DoD’s financial management enterprise and build an efficient system, the Defense Innovation Unit (DIU) and the Joint Artificial Intelligence Center (JAIC) partnered to implement artificial intelligence (AI) and machine learning (ML) solutions.

The DIU and the JAIC integrated intelligent automation into new and existing automated workflows and managed two separate pilots, Summit2Sea and Vertosoft/DataRobot, both powered by AI/ML technologies.

The solution with impact

By integrating auto-ML and AI software with existing RPA infrastructure, the DIU and JAIC have saved millions of labor hours for the DOD.

“In our Humanless Unmatched Transaction (HUnT) solution, we use ML to predict the corrective actions necessary to clear UMTs, and RPA to then perform the necessary clearing entries,” says Erica Thomas, the Comptroller from OUSD Comptroller’s Enterprise Data & Business Performance directorate.

With a better and more efficient system in place now after a 12-month effort, the DOD hopes that its other entities such as the Navy, Air Force, Marines and components will be able to leverage this project in the future for similar use cases.