By Wolfgang Rachbauer, RBI Credit Office, Artificial Intelligence in Loan Documentation

Wolfgang Rachbauer

Transforming a company into an efficient and adaptive organization needs innovative ideas in all aspects of the daily work. Traditional banking is changing significantly. All processes are being analyzed from the scratch and banks are implementing automation procedures even in fields that did not seem to be suitable for automatization. Today, many banks have started partnership programs with fintechs. For instance, Raiffeisen Bank International (RBI) cooperated with Uhura Solutions in the Elevator Lab. Wolfgang Rachbauer, Head of Credit Office in RBI and Use Case Lead in Artificial Intelligence (AI) in the loan documentation process, describes the huge impact of AI in traditional banking.

Artificial Intelligence comprises a wide range of types with various goals, e.g. increase of efficiency, decrease of costs, accelerate processes, etc. There is hardly any industry that will not use AI in future. Especially in the financial industry the demand for AI solutions will increase substantially. Banks will have to become quicker in their decisions. They have to speed up their internal processes in order to increase customer experience. Today, Time2Yes and Time2Cash are crucial KPIs for any bank. AI is one of the keys supporting these developments.

Legal Tech, Natural Language Processing, Deep Learning, Robotic Process Automation – how does this fit in traditional large corporate loan processing? Getting in touch with the fintech Uhura Solutions during RBI´s Fintech Partnership Program “Elevator Lab”, a small team of lawyers, credit specialists and tech experts from RBI´s Credit Office and Uhura´s expert group worked on an interesting hypothesis: Today loan agreements usually have to be analyzed and main conditions need to be manually stored into core banking systems. This process should be automized by training a machine that will be able to read and understand complicated loan agreements “like a human being”. The aspiration was not only to search for paragraphs and phrases with the help of OCR (Optical Character Recognition), but also to read the loan agreement by using AI algorithm.

After training the “Artificial Intelligence Machine” the credit experts together with Uhura technical staff locked themselves up in a testing-center for two days. Like the Mission Control Center in the Apollo 11 lunar project 1969 held the breath during the descent, RBI´s credit team awaited the outcome of the tests. In statistical analysis of binary classification, the F1 score is a measure of a test´s accuracy considering both the precision and the recall of the test, but the setting also comprised various other factors like process run time or end user experience. The outcome was very promising and encouraged the experts to proceed with AI in handling of loan agreements.

“After five months of great experience with Natural Language Processing the experts from Credit Office together with Uhura Solutions were able to present the results to RBI´s Innovation Board. Our experiment was carefully observed by RBI´s Elevator Lab sponsors and finally attracted high attention in the financial community”, Wolfgang Rachbauer explained. “With these results it will be able to substantially speed up the loan process. The first step on the moon is done – now it´s time to prepare for the flight to Mars.” Currently RBI experts are defining the exact set of data in loan agreements that could be extracted by AI and analyzing how alternative processes will look like by using Natural Language Processing.

Uhura Solutions is a company focusing on enhancing processes of financial institutions by using AI technologies such as Natural Language Processing to analyze contracts.