The growing demand for online services and resource-intensive applications has led to the need to optimise the allocation of resources on virtual machines. Traditional methods based on manual allocation are insufficient in the face of the complexity and scale of modern virtualised environments. In this context, it is proposed to develop an architecture based on explainable artificial intelligence algorithms to dynamically optimise resource allocation in virtual machines. The objective is to improve the efficiency and performance of virtualised servers. It is expected to obtain a modular architecture that integrates different techniques of explainable artificial intelligence, contributing to the advancement of knowledge in the management of virtualised systems. The research will provide a solution for the optimisation of resources in virtual machines, improving efficiency and transparency in decision-making.