A 10-point guide to using next-generation AI to enhance the KYC Process
A few months back we published a blog titled “Spotlight on the Remarkable Potential of AI in KYC”. It touched upon how the emerging field of artificial intelligence can help enhance the KYC process, especially in areas like Link Analysis (used for Enhanced Due Diligence) and Pattern Recognition to spot money launderers, financial crimes and questionable transactions.
Overall, the term AI has been around for a while. A quick Wikipedia search reveals that the field of modern AI was born in the year 1950 when Alan Turing published a seminal paper on building machines that can think. Here we are, almost seven decades later still in the early days of this technology.
Over the last few years, Google’s CEO Sundar Pichai has been speaking about the increasingly crucial role of AI in software and it seems like this year might just be the inflection point for the field. In May 2017, Pichai explained how at Google it is an “AI first” approach for several of its products. From mobile-first we now have a new phrase that has the potential to impact many different sectors. Tesla’s Elon Musk has launched the OpenAI project to conduct cutting edge research on the subject and Google has unveiled AutoML, a platform to build AI tools using AI itself – a potentially crazy yet real suggestion.
In the world of banking and financial services, technology adoption has always been quick. In the next few years, we expect an “AI-first” approach in several financial transactions and banking processes. At Cenza, as always, we want to stay ahead of the curve. Our leaders are evaluating the role AI can potentially play in enhancing our ability to serve our banking and FI customers better.
In this post, we unveil a 10-point guide to how AI and machine learning can possibly impact KYC/AML methodologies adopted by banks and financial institutions.
1. We believe one area where AI and Machine Learning can really impact the KYC process is in helping identify customers who need to be screened with an Enhanced Due Diligence (EDD) process. Based on pattern recognition techniques coupled with algorithms to analyse unstructured data, it would make it much more efficient to identify relevant candidates for EDD. Moreover, there is a critical problem of ‘false positives’ which must be investigated and eliminated. A machine learning process that is able to unearth and reduce false positives can be a game changer in the KYC space.
2. Even today, AI and bots are really useful to perform repetitive tasks. Using chat bots to ask customers questions and, analysing their responses using NLP, can certainly save critical time and staffing needs to run a KYC process. For example, ML algorithms that crawl through social media data to perform risk-based assessments, can be included in the KYC process.
3. Emerging techniques to run NLP (Natural Language Processing) algorithms to analyse data from voice and text can help banks truly get to know their customers. Banks and FIs would do well to go beyond the usual sources of customer data, and using NLP, will have the ability to crawl through more sources and documents.
4. One of the biggest challenges in KYC is keeping up with regulation. If your KYC Toolkit can bring to the fore AI-enabled features to keep track of constant regulatory changes and suggest action items, it would save critical time and effort. Some AI are even trained to understand ever evolving regulatory changes, identify gaps in collecting information and generate alerts for process completion.
5. One of AI’s biggest advantages will revolve around delivering workflow automation. If automated processes can be established to spot moves by competition, keep track of emerging regulations and enhance productivity of KYC staff, it would bring in serious efficiency into the KYC process.
6. Automation of SAR filings, report generation and visualization techniques to make sense large volume of unstructured data, can all be delivering using simple ML algorithms.
7. As we wrote in our earlier blog post, one of the biggest challenges in the KYC process is making sense of a complex web of transactions. Link Analysis, as the process of connecting various transactions leading to the UBO, is currently a complex process. Automation could be a game changer in this area.
8. As the very definition of AI suggests, a bank’s KYC toolkit must “constantly learn”. Be it pattern recognition modules or NLP to crawl through millions of pages, the KYC framework used must “learn” continuously.
9. Cross-enterprise compliance, across a FI or bank’s various geographies, is currently a challenge. An AI-powered automated workflow would make it seamless to deliver enterprise-wide systems and processes.
10. Last but not the least, KYC is all about truly getting to know a customer, his or her risk profile and putting them into relevant customer buckets for due diligence efforts. It is also about keeping up with regulation to avoid the big penalties as institutions have learnt the hard way. AI techniques, especially around Bayesian learning frameworks can help build more accurate risk scores for all customers.
This 10-point guide to the application of AI methodologies in the KYC process is by no means an exhaustive list. At best, it’s a guideline to help banks and FIs pursue automation techniques with all seriousness.
At Cenza Technologies, we constantly upgrade our KYC Process to include the latest of technologies to serve our customers. With a deep customer-centric approach and cutting edge technology, we’re able to run a robust KYC/AML services practice for banks and financial institutions that are amongst the best in the world. Do reach out to us to know more.