Design and Implementation of an Intelligent Call Routing System in the Banking Sector for Improved Customer Experience

The banking sector is undergoing a progressive transformation of customer service processes driven by artificial intelligence, automation, and increasingly sophisticated data-driven decision-making systems. Within this context, contact centers represent a significant opportunity for digital transformation, particularly in the way incoming customer requests are classified and routed to the most appropriate operator. This thesis investigates the design and evaluation of an agentic artificial intelligence architecture for intelligent intent classification and operator routing in a banking contact center. The proposed solution combines customer and CRM information with conversational evidence to construct a structured representation of the customer context, which is then processed through a coordinated agentic pipeline. The architecture uses an LLM to identify the most likely customer intent, a deterministic confidence-scoring mechanism to assess the strength and consistency of the available evidence, and a routing component to match the predicted intent with the skills and characteristics of available operators. The system is developed and evaluated as a proof of concept using synthetic CRM and call data. Rather than presenting the results as evidence of production level performance, the study uses them to assess the architectural potential of the proposed approach and to identify the conditions required for meaningful realworld validation. Beyond the technical implementation, the thesis assesses the potential business value of intelligent routing from economic, customer-experience, and organisational perspectives. Overall, the thesis demonstrates how an agentic AI architecture can provide a structured foundation for intelligent contact-center routing in banking, while emphasising that its effective adoption requires not only technical feasibility but also empirical validation, human oversight, governance, and continuous improvement.