Blog

The merits of artificial intelligence in the customer due diligence process

Sunny Nagpal Senior Manager, Banking and Financial Services Solutions
Print This Page

March 28, 2017 - Financial institutions (FIs) continually strive to earn the trust of their customers, but equally important is being able to trust and verify the information customers provide back to them. Regulators across the globe are increasingly focusing on ensuring that banks have robust and effective controls in place for customer due diligence (CDD). Lapses in money laundering and customer due diligence controls exposed in several FIs have resulted in regulators tightening their levels of supervision on the industry as a whole.

A few recent examples:

  • Standard Chartered ran afoul of sanctions and agreed to a $327 million settlement with US law enforcement agencies
  • BNP Paribas was charged with sanctions violations and agreed to pay $8.9 billion in fines
  • Others, such as J.P Morgan Chase, Citigroup Inc., and Wachovia Corp have also been cited for anti-money laundering lapses or sanctions violations

For FIs, customer due diligence is an important element to manage risks and protect themselves against potential financial crimes.

Facing the challenges

The Know-Your-Customer (KYC) process poses a series of challenges for banks that must be taken into account before an effective customer due diligence program, aided by artificial intelligence (AI), can be implemented:

  • Non-standardized regulation: Regulations and regulatory environments across the globe vary considerably. Multinational banks need to ensure compliance not only in their home country, but also in environments that are more complex and have less infrastructure.
  • Insufficiently trained resources: KYC for institutional and corporate clients as conducted in most large banks is heavily managed out of large operations teams situated in various parts of the world. These large teams are not always trained to effectively identify and flag suspicious activity.
  • Strained client relationship and delayed transactions: The greater level of complexity and detail of information required leads to longer onboarding and renewal times, straining client relationships and delaying revenue-generating transactions.
  • Legacy systems for document and workflow management: While KYC documentation is typically collected and retained over several years in document management systems, these documents are not always easy to find—for instance, there are often multiple documents for each entity under a single document type. Also, migration to more modern workflow management tools is slow and expensive.

It is imperative that FIs, given these challenges, look at implementing KYC, CDD, and overarching anti-money laundering (AML) procedures in a way that is not only cost and time effective, but also capable of identifying and appropriately flagging risks.

Moving ahead in a "digital-first" world

There are several solutions available to address the problems associated with predominantly manual CDD tasks and activities. The solutions range from basic screen scraping-based repetitive-task automation to enhanced AI based solutions. In the case of the CDD process, many complex activities cannot be simply automated, as the process is subjective and uses a variety of inputs. However, as overall AML and KYC processes continue to digitize, the ability to use effective AI solutions will increase. AI-based solutions are capable of simulating human intelligence, including complex decision-making through the evolution of Machine Intelligence. Machine Intelligence is the use of AI and deep reasoning to accomplish work. It includes Machine Learning, which is used in cases that are today done manually in the CDD process, such as exception handling and management.

Designing an intelligent CDD solutions

One example of a more intelligent CDD solution is Genpact's neural KYC process, which leverages cognitive technologies and is based on a standardized AI platform. By incorporating tools such as a language neutrality engine along with Machine Learning, relevant information from documents and systems can be easily extracted in an automated fashion. By then producing the information in a synthesized format for experts trained in the nuances of AML and KYC, the starting line for CDD is immediately moved forward. In addition, enabling the automatic update of information sourced through systemic mining (after analyst verification) takes the ability to update correct documents within various systems one step farther without requiring additional resources.

While this is just one example, there are multiple approaches that can be taken to bring an AI-based solution to life. In some cases, the best option could be to start with an end-to-end AI platform and leverage parts of it that are relevant to building the solution; in others, all that might be required is to stitch existing technology components together in order to architect the solution. Which approach to opt for depends on the business and operating models of a particular FI, taking into account criteria such as scalability, flexibility, and other aspects of integration. However, the most important thing to consider is that for any AI based solution, it is not just about finding the technological components of the solution, but also knowing ahead of time how those components will work together as part of an architected CDD solution. 

Continue Reading

Ready to
Explore?