Suman Singh, Founder and CEO, CyborgIntell runs an AI-ML-enabled data analytics platform useful for the MSMEs. Here he talks about his work.
How successful is an AI-ML based data analytics platform in helping business growth of MSMEs?
Suman Singh - The data generated within businesses has become one of the most important elements responsible for their growth journey. However, as per a Gartner report, nearly 97 per cent of data sits unused by organizations. Data, if used wisely by aligning the business objectives and the technology, surely can help tackle one problem at a time and create robust solutions to some of their business problems. In India, businesses are still struggling to integrate these technologies into their operations. They are still largely dependent on human talent and conventional resources to make decisions based on these insights across all the sectors including BFSI. It is this gap that prompted CyborgIntell to explore the BFSI industry and work towards improving the adoption of AI and ML and ensuring safety from financial fraud.
A few of the use cases of AI in the financial sector-
- 28% Increase in Loan Approval Rate using AI-powered Automated Credit Risk Approval: Determine risks at an individual level through AI-based credit decisions. Enable straight-through processing and rejection, drive efficiency in approvals and create Dynamic Scorecards using ML-based predictions
- 40% reduction in Customer Acquisition Cost for lending companies using AI to optimise the customer journey, Deep customer segmentation to determine good customers, Optimize sourcing and drive effective marketing campaigns.
- 216% Lift in conversions on Marketing Campaigns to drive Up-sell/Cross-sell Opportunities with a focus on the right set of leads for conversion, optimize sales efforts and increase product penetration
- 39% Increase in collection efficiencies using AI-driven Delinquency prediction for various buckets, predicting high, medium and low-risk defaulters within each bucket and creating custom contact strategy based on risk severity score to optimize debt collection.
How cost-intensive is an AI-ML-based data analytics platform for the MSMEs in non-metro cities?
Suman Singh - For solving any data-driven business problems, financial institutions require multi-disciplinary skills: Subject Matter Expert, Data Scientist, Data Engineer, and Develops Engineer to develop and integrate prediction & intelligence with their Loan Origination System and Loan Management System. Hiring multiple resources, waiting for many months to go live and if anything goes wrong then re-do the entire work are very costly affairs for any organisation.
What is the geographical reach of your platform across India, in metro and non-metro locations?
Suman Singh - MSME business is booming right now, lenders are providing credit facilities to Kirana stores, MOM & POP shops, and many other businesses in tier 1 & tier 2 cities. Our existing customers are using our platform to assess credit worthiness of MSME customers and are able to offer the right credit lines.
What is the rate of accuracy in AI-ML-based predictive analysis?
Suman Singh - Let me explain this with the help of a use case that we did for one of our clients. They wanted to Improve Collections and optimize efforts. So the challenge here was - The NBFC sector has undergone a significant transformation over the past few years and plays a significant role in the growth of any financial system. NBFCs have outperformed banks across several product lines by carving out niche products. They are more customer-focused and take the time to understand customer behaviour and build customised products and reach out to different segments of customers with customised loans and customer-friendly repayment plans and take higher risks. This however creates the challenges of debt collection. Debt collection is important for the company to improve their cash flow and help businesses reduce the risks of incurring losses, and free up their resources.
And the solution that we came up with is - iTuring can be used to develop predictive models that identify customer default early in their lending journey. It can accurately forecast delinquency movement for the whole portfolio, across all customers and all buckets. The outputs of the default prediction models and their explanations around customer behaviour can help define strategies to improve overall collection efforts and as a result, improve the portfolio. The FinTech company we engaged with on Collection optimization was experiencing a default rate of ~4.2%. We used iTuring to build predictive models that predicted customer movements from one delinquency bucket to the next for pre-delinquency, early stage, late stage and recovery. iTuring developed accurate models which predicted default in the immediate next month with an accuracy of ~86%, enabling businesses to effectively manage portfolios monthly. This enabled the company to identify 10 customer segments based on the probability of default and value at risk and develop collection strategies around the same. By focusing efforts on 82% of likely defaulters that were identified in the top 30% of customers the company would improve collections efficiency by 39%.