Voice and Text Assistant
Contact Center Automation
Client and Industry
Our client is the contact center of a major financial company with an active customer base of over 10 million.
As the business grew, the challenge was to maintain a stable revenue level without increasing the number of operators. The goal was to automate routine inquiries while maintaining service quality and customer loyalty.
Project Goals and Objectives
The primary goal was to implement a voice and text assistant capable of handling routine customer inquiries without operator involvement. At the same time, it was essential to maintain a high level of trust in the brand. The assistant needed to work predictively — understanding the context of a request by analyzing transactions, ATM activity, and the customer’s typical behavior. Human operators remained responsible for handling complex or expert-level issues, but their work was also enhanced with a system of real-time prompts and recommendations during conversations.
Work Completed (Services Provided)
Achieved Results
In the first months of the assistant’s operation, the number of voice inquiries decreased by more than 50%, while text inquiries in chat dropped by over 70%. Additionally, in 30% of cases the system helped operators speed up customer service through prompts and recommendations, reducing the overall request handling time.
Technologies and Tools Used
A Business Rules Management System was implemented, along with an interface for managing the voice and text assistant. Integration with the bank’s CRM and telephony systems was also completed.
Conclusions
The project demonstrated that automation can be seamless and customer-friendly. Through predictive analysis of customer interactions across the bank’s touchpoints, the system was able to anticipate questions and reduce the workload on the contact center. Even when a customer was transferred to a human operator, the automation continued to work in the background, enhancing the employee’s expertise.
During the implementation process, various speech recognition and natural language processing technologies were tested. Since it is critical for a bank to provide accurate information, a model was chosen that combines AI trained on regulatory documentation with elements of manual control in sensitive scenarios. As a result, a high level of automation was achieved without compromising service quality or the brand’s reputation.