The credit lending industry has witnessed a strong upswing over the last few years. According to CRIF High Mark’s report titled ‘How India Lends’, India’s lending market grew to Rs 174.3 lakh crore in March 2022, up by 11.1 percent year on year.
Several players have emerged in this space to cater to the new-age digital consumer. Leading the pack is RING, a transactional credit app launched by Kissht.
Ranvir Singh, Founder & CEO, Kissht | RING; Krishnan Vishwanathan, Founder & Executive Director, Kissht | RING; Karan Mehta, Founder & CTO, Kissht | RING and Sonali Jindal, Founder & COO, Kissht | RING are at the forefront of building solutions offering instant online credit all in one click.
Karan and Sonali spoke about the vision and mission of their startup, the challenges they have faced along the way, and how their collaboration with AWS has helped them scale and innovate.
The mission of RING
At RING, the mission is to reach out to the section that is yet to receive the benefits of digital financial services.
“When we look at the penetration in this segment, whether financial services are being offered or not – it is as low as 5 percent. That’s why we want to cover these 150 million millennials through a simplified product offering, which is a combination of payments and credit,” shared Sonali, adding that their idea is to provide a simple, analytically-led digital experience.
All it takes is less than a minute for a customer to download the app and get the loan.
The journey from then to now
Sonali and Karan started using AI/ML solutions for their lending platform Kissht and carried those solutions forward to RING.
In their initial days, Kissht was focusing on a traditional underwriting model, where they were looking at a selected set of data. Over the last five years, they have served close to around 5 million customers and disbursed loans to 3 million customers.
“Given the data of customers that we have, all our underwriting models at RING are now geared towards using artificial intelligence and machine learning, in terms of our day-to-day decision-making. For us, technology and analytics run across both our core and non-core offerings and that is where we bring a big difference to the table,” explained Sonali.
For RING, the biggest challenge was to tap into a large customer set that didn’t have a prior CIBIL background. In that case, how did they underwrite and assess customers who didn’t have any credit history?
Karan answers, “The good thing is that today we have a rich digital footprint of customers. Most of our transactions are online, which is enough to assess a customer even if they have not taken credit before applying to RING.”
Today, their underwriting and fraud models are governed by machine learning. In the beginning, it was all a little daunting, where they assumed that a large team of experts would be needed. Over time, their exposure to a host of services through Amazon made the journey far more seamless.
Leveraging the power of AWS
RING relies on Amazon Sagemaker in a significant way. When they started using the service, they realised that a large part of the heavy lifting was already done. Most of the time, training machines start much later – the initial few months are spent setting up an infrastructure, but none of that had to be done here.
“Our credit and fraud engines make use of Amazon Sagemaker in a big way. We have teams that are building, testing, and deploying models, all using Sagemaker and Sagemaker Studio, which is fulfilling our primary purpose. RING‘s credit engine has been trained on over 14 million loans, and the fraud engine has verified 20 million customers. We can differentiate a risk accurately and can predict in advance how a certain month’s performance is going to be, at the time of disbursing,” explained Karan.
To be able to extract printed text from crucial documents, the team at RING leverages Amazon Textract. This helps to automatically extract printed text, handwriting, and data from any document. While there were ready models available for Aadhar and PAN Card, the other documents were processed using this service.
AWS Rekognition is yet another cutting-edge solution that helps the RING to identify customers and their photos.
“We have use cases that allow us to do 1:N face dedupe, blur detection, or just making sure that the photo is accurate, not a photocopy. We also ensure that it is not a photo of a device screen,” added Karan.
Currently, they are testing more use cases that involve voice detection to understand the customer’s tone, when they call the support team.
Business Impact
There are five areas where AWS has had an impact on the business, revealed Sonali. For one, it has reduced NPAs by 20 percent to 25 percent. The second is from a retention standpoint, where RING has scored upwards of 90 percent cycle-on-cycle for 16-18 months.
“When I look at customer engagement, or customer service, our ability to say that these are the set of customers I should service and these are the customers who can wait is a differentiator. Moreover, the wait times in terms of customer service have gone down by 50 percent,” she shared, adding that the collection delinquency has also reduced by 20-25 percent.
The road ahead…
RING’s vision is to provide an app that is fast and well-designed, and at the same time, shows the right options at the right time.
“If a customer is outside a store, show him a QR code scanner. If he is at home and shopping online, show him the right online payment options. To be able to achieve those use cases, you need ML,” concluded Karan.
