KYC data cleansing projects can be avoided with artificial intelligence

Increased international compliance requirements and a lack of investment in this area have made large-scale KYC data cleansing projects necessary. But do these projects really solve the quality issue? Not in the medium and long term. In addition to the initial investment, continuous efforts are also required to maintain the quality of KYC information and thus to ensure that another cleansing project is not just around the corner.

How can artificial intelligence (AI) help?

In addition to a bank's cultural and process considerations, the allocation of financial resources is one of the key issues in ensuring consistently high-quality KYC data, not least due to the cost of the compliance department and experts involved. Annual reviews to ensure continuous quality assurance make up a large part of this cost. Therefore, it makes sense to consider using systems based on the use of artificial intelligence that may be able to implement quality assurance measures in a more cost-effective manner.

Automation – a sensible intermediate step

Robotic process automation is one technology that is already widely used to improve efficiency. Software robots help to fully or partially eliminate repetitive tasks for compliance employees, and a typical application is the automatic compilation of open source information. However, the compliance decision of whether KYC meets a bank’s requirements is still made by an employee. Robots are thus particularly useful for the preliminary work.

Natural language processing – the next step

Using a computer to determine the quality of all or part of the content of KYC information seems an obvious choice and the next logical step. Artificial intelligence has taken off in recent years due to increased processing power, so it could very well make sense to consider this possibility. Since KYC files contain a large quantity of unstructured information, natural language processing (NLP) has a lot to offer. Entire texts are vectorised in the form of complex matrices and then mapped in a mathematical model. Previous evaluations by a compliance officer and the underlying KYC data are fed in during a training phase to build a model in several iterations. Following this, not only can the quality of a KYC file be determined immediately, suggestions for improvement can also be generated for a relationship manager. Enormous savings can be made as a result – whether during client on-boarding or the annual KYC review.

The compliance officer of the future

Depending on the strategy followed, a compliance officer can now consider whether the efficiency gains should be used for other compliance activities, or whether a contribution should be made to the cost targets. What is certain is that the role of a compliance officer will change for the better, thanks to the elimination of repetitive, error-prone tasks in favour of varied, investigative activities. These new capabilities also represent an interesting addition to this field of work in dealing with AI-based systems. This will have no bearing on the importance of a compliance officer in the provision of financial services. Quite the opposite: the compliance officer will remain essential in order to prevent financial crime and minimise risk.