Introduction to AI in Financial Services
The integration of artificial intelligence (AI) in financial services is moving towards increasingly specialized models. In the realm of credit risk management, tailored solutions have demonstrated significant advantages over general-purpose models like ChatGPT or Gemini, according to Yoel Gavlovski, founder and CEO of Quash, a company specializing in financial analysis systems.
Comparative Testing Results
Quash conducted comparative tests where a specialized risk credit agent was evaluated alongside a general-purpose AI model. The results showed that the specialized system could generate artificial information mimicking real data, request additional details to refine estimations, and produce expert-validated analyses. In contrast, the generalist model exhibited conceptual errors and technical limitations when addressing financial domain tasks.
Benefits of Specialized AI in Credit Risk Management
Regulatory Compliance Assistance:
“Generalist models aren’t designed to comprehend the complexities of credit risk. Specialized AI can automate data processing, generate intelligent variables, build predictive models, and ensure regulatory compliance more efficiently. In essence, it transforms highly technical processes into accessible and reliable tools for financial decision-making,” Gavlovski explained.
Potential Impact on Banking Sector
According to McKinsey & Company’s study, “Capturing the full value of generative AI in banking,” implementing specialized AI could boost operational profits by 9-15% for the banking sector due to efficiency improvements, cost reductions, and optimized critical processes like risk management and customer service.
Real-world Application Example
Gavlovski highlighted the potential of these tools using a case study of a sportswear retail chain with over 150 branches seeking to provide financing to young individuals without credit history. By leveraging existing customer data and specialized AI, they identified patterns and extended credit to individuals with no credit history in some credit bureaus.
This approach broadens credit access based on user behavior similarities, applicable also to e-commerce. If a applicant lacks credit history but shares traits with 300 other on-time payers, the model can infer a high compliance probability. Online shopping and financial behavior have become strong predictors of loan repayment willingness, enabling over 70% accuracy in distinguishing good or bad payers, particularly in Mexico’s financial inclusion segments.
Adoption of AI in Fintech
Approximately 68% of fintechs already use AI in their operations, with many relying on external providers, as per the “Fintech Radar 2025” report. The choice between internal capabilities and third-party solutions largely depends on the function being automated or enhanced.
- High Adoption Segments:
- Wealth management: 81% of companies reported implementation
- Digital banking: 73%
- Technology infrastructure for banks and fintechs: 69%
