In multi-agent systems (MAS), the task delegation process involves assigning tasks or responsibilities among the agents to optimize the system’s overall performance. Nevertheless, choosing which agent must receive a specific task or responsibility can be seen as an exploitation/exploration problem. In the Multi-armed Bandits (MAB) context, such a problem can be solved by deciding which agents (partners) should be exploited or explored concerning a certain task. However, for a system composed of several agents, it is important that such a choice be made based on a delegation model capable of providing mechanisms to explain the agents’ decision-making process, making the partner selection more transparent and understandable. This feature is especially desirable when different partners are available to execute a task, resulting in a large set of combinations of partners and tasks. In this paper, we introduce a Multi-armed Bandit (MAB)-based delegation model that can optimize the partner selection process and explain the agents’ choices. Our approach allows agents to explain their partner choices based on quantitative argumentation with votes (QuAD-V). We validate our model by simulating petroleum product distribution via pipelines, where agents represent temporary storage bases in a complex delegation chain. The results demonstrate the effectiveness of our model in optimizing delegation decisions while maintaining clear, understandable explanations for agents’ decisions.