Toward an Inclusive Cognitive Banking System: Empathetic Voice Assistants for the Digital Autonomy of Elderly and Visually Impaired People in Africa

Abstract

The digital transformation of the banking sector has significantly improved access to financial services and operational efficiency. However, it also reveals a persistent usability gap, particularly affecting elderly and visually impaired individuals who often struggle with complex digital interfaces. This paper introduces an inclusive cognitive banking approach based on an empathetic voice assistant, enabling users to access financial information through a simple phone call without relying on graphical interfaces. The proposed architecture integrates GSM telephony, an Asterisk-based call management system, realtime audio processing through LiveKit, and a conversational agent adapted to users’ cognitive abilities. A key contribution lies in the integration of the Model Context Protocol (MCP) as a secure intermediary layer between the intelligent agent and the banking database, ensuring controlled access, structured interactions, and full traceability of requests. Experimental results confirm the feasibility of the system, highlighting intuitive voice interaction, robust speech understanding, and a secure framework for accessing sensitive financial data. The study demonstrates that combining telephony, voice-based artificial intelligence, and MCP provides an effective solution to reduce barriers to financial services in constrained environments. Future work includes multilingual system extension, real-world large-scale evaluation, and strengthening MCP-based security mechanisms.