Delivered an AI-assisted loan approval platform

Category

Case Study

Author

Published

May 26, 2024

Introduction

Our client faced significant challenges in their loan approval process, including information gaps leading to denials, high rates of willful defaulters among new customers, fragmented KYC data, and dissatisfaction among affluent, influential customers. To address these issues, the client sought a solution to enhance their loan approval process and improve customer satisfaction.

Analyzing the Problem

  • Loan denials stemmed from insufficient applicant information, contributing to delays and customer dissatisfaction.
  • Historical data revealed a pattern of new customers being the primary defaulters, highlighting the need for improved risk assessment.
  • Fragmented KYC data across various systems hindered comprehensive customer evaluation and risk assessment.
  • Denials of loans to affluent, influential customers led to dissatisfaction and potential reputational risks.

Initial Challenges

  • Consolidating disparate data sources and silos posed challenges in gaining a holistic view of applicants.
  • Identifying and mitigating risks associated with new customers and willful defaulters required advanced predictive modeling.
  • Effectively segmenting and analyzing customer data to address the needs of different customer segments was essential for success.

Our Solution

  • Leveraging a multi-modal model incorporating tabular, text, and image data enabled comprehensive applicant evaluation.
  • Implementing XGBoost for predictions and Vision Transformers for data extraction enhanced accuracy and efficiency in risk assessment.
  • Utilizing comprehensive social media analytics provided valuable insights into customer behavior and creditworthiness.

Key Results Achieved

  • Implementation of the AI-assisted loan approval platform led to a 2x increase in customer satisfaction scores, improving overall customer experience.
  • The optimized loan approval process resulted in a 2x growth in loans disbursed, expanding the client's lending portfolio.
  • The AI-enabled platform facilitated better risk assessment, leading to a 5x increase in the profitable customer base.
  • The AI-assisted loan approval platform streamlined operations, reducing processing times and enhancing productivity.

Conclusion

By implementing a multi-modal framework model, advanced predictive modeling techniques, and comprehensive social media analytics, our solution successfully addressed the challenges faced by clients in their loan approval process. The project resulted in significant improvements in customer satisfaction, loan disbursal, and profitability, laying the groundwork for continued success and growth in the competitive lending market.