 This study examines the feasibility of using a river as part of a last-mile delivery system. Two different network designs were proposed, one involving a barge and the other involving e-cargo bikes. The efficiency of these scenarios was measured against various performance indicators. A three-stage decomposition heuristic was used to solve the problem, with the first stage involving a non-supervised machine-learning clustering algorithm to determine which parcels should be assigned to the closest satellite. The remaining two stages involved routing decisions for both the barge and e-cargo bikes. Results showed that the use of a river transportation mode and e-cargo bikes resulted in a 41% reduction in fixed costs and a 92% reduction in energy consumption compared to traditional trucks. Future research should consider additional costs such as fuel and maintenance, as well as a larger network size. This article was authored by Angie Ramirez Villamil, Hiro Arman Toyotores and Anisha Jagler. We are article.tv, links in the description below.