Conversational Commerce Architecture Using MCP-Enabled AI Assistants for Multi-Marketplace Price Aggregation in Emerging African eCommerce Markets

The proliferation of mobile internet services in Sub-Saharan Africa has led to the emergence of a fragmented eCommerce ecosystem with consumers required to switch among different marketplaces in order to identify the best deals. In this paper, we examine an MCP-driven conversational commerce solution known as Shop Africa / Shop Everywhere (SASE), which integrates real-time product information from different eCommerce websites in Africa such as Jumia, Jiji, and Konga, into an easy-to-use chatbot interface. We propose a five-level architecture of the SASE solution consisting of NLP, MCP orchestration, marketplace adapter, data normalization, and presentation layers. Moreover, we develop a Formal Price Dispersion Model (PDM) to quantify potential consumer savings resulting from cross-platform shopping and a Query Response Latency (QRL) model to evaluate system performance in a high-user load scenario.
As demonstrated by our demo, the SASE framework achieves a mean price dispersion index of 0.21 in those three websites. Customers will be able to save more than 22% in their purchases of popular electronics through comparison shopping. The outcomes illustrate the viability of MCP-based commerce aggregation from an architectural point of view. This provides a basis for extending the architecture to international online marketplaces such as Amazon, eBay, and Best Buy. It also carries great implications regarding digital inclusion and consumer welfare.