Sella cuts credit rating inquiry response time by 99% per case with Prometheux

Sella's credit rating team computes monthly credit scores for all clients, feeding a risk framework for provisioning, compliance and decision-making. Rating changes trigger inquiries from branches, and analysts had to manually retrieve data, compare outputs and write precise explanations, taking up to 17 minutes per case, one case at a time, in complex Excel workflows. With millions of clients, the manual analysis became unsustainable.

Sella's credit risk team used Prometheux to connect to customer data, translate convoluted Excel workflows into simple logic and connect to LLMs for explanations in natural language. The system accesses Sella's databases, processes each client's historical variables and provides an explanation autonomously. It automatically identifies critical variables and generates explanatory comments in natural language, as well as what-if simulations, with end-to-end data lineage. Sella also gets automated summary reports for branch managers and a direct database connection.

Prometheux's case study reports that Sella cut response time by 99% per case on average, from up to 17 minutes per case to 5 seconds, and says the system scales to millions of analyses in parallel. What-if scenarios are now analyzed instantly and proactively rather than through manual Excel recalculations. A credit risk analyst said the team used to spend hours doing this by hand.

• Ground LLM explanations in explainable data processing with data lineage so credit teams can trust and audit the output.
• Translate the logic analysts keep in Excel workflows into simple, executable logic and connect LLMs to it for natural-language explanations.
• Personalize the explanations: Sella's analysts say generic text would undermine credibility.