More than a billion of the rural merchants in the developing world commonly depend on hiring on-demand transportation services to commute people or goods to markets. The process of selecting the optimal fare involves handling decision-making characterised with multiple alternatives and competing criteria. Decision support systems are commonly used to solve these types of problems. However, most widely used systems are based on object-based approaches which lack high-level abstractions needed to effectively model and scale human-machine communication. This paper reviews previous literature on the field and introduces an improved preliminary agent-based decision-support approach to overcome those challenges. As a proof of concept, we developed a two-agent simulation that, given a request from one of the agents, the other one takes a dataset of a stratified sample of 104 Ethiopian commuter criteria preferences taken from the Dukem region and an exemplary dataset of fare alternatives. The assistant agent computes those datasets using widely used HPA and TOPSIS algorithms to weight, score, rank those alternatives. Once we run the simulation, in a matter of milliseconds the assistant agent effectively returns an optimal prescription to the other agent, storing all interactions in a self-contained memory resulting in an architecture that allows developers to program further customisation as interactions scale.