Contactless Lending Would Be Powered By AI For Better Credit Decisions And Personalized Experiences: R P Singh, CEO, Nucleus Software
Governments all over the world are taking extraordinary measures to improve the economic conditions. The role of financial institutions in deploying these measures effectively and efficiently calls for accelerated digital capabilities to be brought to bear.
The post-crisis regulatory frameworks have been gradually settling into place, and financial institutions have been adjusting their business models accordingly. Digitization is not just about mobile, it is about being digital from the first interaction with customers all the way through – that includes channels, middle office, back office and indeed a full set of capabilities such as APIs, workflows and engines.
Nucleus Software claims to have partnered with many financial services firms enabling them to offer end-to-end digital lending experiences, using AI and analytics to help in making better credit decisions faster, offering personalized experiences, and mitigating risk.
In an exclusive interaction with BW BusinessWorld, Nucleus Software CEO, RP Singh outlines the ways in which technology has accelerated many aspects of digital banking during the pandemic. Excerpts:
Much of the government stimulus for COVID-19 has been in the form of credit access, particularly for the MSME sector. How can new-age technologies help increase credit access to MSMEs?
MSME borrowers have a few challenges such as their inability to meet stringent credit criteria set by lenders, complicated and time-consuming processes at lenders, lenders finding it difficult to ascertain their risk quotient, and many more. Today, these challenges can be addressed quite effectively with the right set of technologies.
Analytics & AI/ML leveraging alternate data sources can be used to build sophisticated scorecards for comprehensive credit assessment, while APIs can seamlessly integrate with third-party data sources and fintechs to provide inputs for assessing customer’s capacity to pay. However, while digital channels have brought financial institutions closer to borrowers than ever before, what we need is the right mix of a virtual and physical approach since it is difficult to develop a deep relationship for financial services in a “digital-only” world. The lenders that get this ‘phygital’ approach right will emerge as the winners in times to come.
And this means that no matter what channel the customer chooses, the lender must be able to deliver an experience that is consistent and seamless. Contactless lending – from acquisition to servicing – will be table stakes in the new world.
From a financial perspective, lenders need to mitigate the immediate challenges posed by the pandemic and adjust to the short- and medium-term economic implications. For example, early identification and management of stressed borrowers – hopefully early enough to take preventative action, offer innovative restructuring options and, where appropriate, managing collections and recovery effectively.
How has the pandemic affected the digital transformation journey at financial services firms? What would be the focus areas hereon?
The pandemic has had the effect of fast-tracking many of the digital initiatives. They are focusing on three key horizons - to react immediately, adapt quickly and to lead after that. Lenders should plan on coming out of the crisis stronger, more agile, and more competitive than before.
As the situation improves, the focus will start shifting towards realigning business models to new realities, re-designing processes and shifting gears for growth opportunities, including re-evaluating the lines of businesses.
What’s your play here?
Nucleus recognized the need for systems to be truly digital – digital end to end –and that’s how we architected FinnOne Neo - an integrated and comprehensively digital platform which delivers digital experiences across various channels through the entire lifecycle of the loan. We created FinnOne Neo by combining the power of advanced technologies with our deep understanding of financial services. For three decades we have been helping some of the world’s most innovative banks and other financial institutions such as Bank of Queensland, CIMB Bank, CRDB Bank to name a few, transform their operations globally. We have gained over 50 customers for FinnOne Neo Cloud which includes: Muthoot Fincorp, RattanIndia Finance, Roha Housing Finance, Esskay Fincorp, Sai Point Finance and many more.
Digital payments data are an important source of data which gives a good measure of the intensity of business, allowing businesses with a low asset base or inadequate credit history to avail timely funds. What are the possibilities with this rich source of data?
As more people and businesses rely on digital payment platforms, the wealth of data generated by those transactions has become increasingly valuable. However, the real challenge is how to turn that data into actionable insight – how to use it to discern credit worthiness. In the past complex but static risk models were used but they are no longer fit for purpose.
AI and ML crunching big data offer tremendous promise but a key challenge with most AI systems is that they are “black boxes” – which means that we do not know why they came to a particular decision. So, if the AI declined a loan application the lender wouldn’t know why and wouldn’t be able to explain it to the applicant. This lack of transparency is very problematic. The good news is that we now have white box AI solutions coming on stream, that explains how they behave, how they produce predictions and what the influencing variables are.
We also created our lending analytics solution based on ‘white box AI’. But, as with all new technologies, care must be taken. AI learns by reading the data it is provided, and if the data is biased then the learning will be biased. Biases are not always a problem but when making credit decisions it is vital that the biases are clearly understood. In my opinion, we should definitely leverage alternate data sources, but we should start by using more explainable statistical-based analytics whilst testing AI/ML decision in parallel to further build confidence.
