OTP Bank accelerated the analysis of customer requests 30 times using Yandex B2B Tech neural networks

2025: Yandex B2B Tech Technology Testing

OTP Bank tested Yandex B2B Tech technologies to improve the quality of customer service. Thanks to this project, the speed of processing and analyzing customer messages has increased 30 times. The bank announced this on August 4, 2025.

Customer support OTP Bank processes more than 400 thousand text requests on 260 topics per month. Analyzing such a volume of information takes a lot of time for specialists, so the bank automated this process using the voice analytics service Yandex SpeechSense. So, during the pilot, more than 29 thousand dialogues with clients were analyzed in 7 days - this is 30 times faster than with the manual work of three employees.

Communication analysis tools, such as Yandex SpeechSense, helped OTP Bank improve the quality of support work. The system automatically determines the topic of cases, distributes them into groups and selects significant semantic blocks in dialogues. This approach helped the bank identify areas for improving customer experience.

The second pilot project aimed to automate client consultations using a YandexGPT-based AI agent. The agent responded to customer requests in mobile app chats on 30 popular topics, such as partial early repayment of a loan, clarification of the terms of the grace period and others. The bot actually answered 60 thousand requests without errors, in connection with which the customer satisfaction index (CSI) rose by 0.2 out of 10 points. In addition, the level of automation of dialog processing increased to 8%. The AI agent worked best in scenarios where the client needed to give a detailed step-by-step explanation or clarify previously received information. The company plans to expand the use of neural networks for processing more complex requests and system work with customer feedback.