The analysis and simulation of problems associated with reservoir geomechanics are traditionally performed using the finite element method. However, to perform this analysis, it is necessary to consider a region much larger than the region in which the reservoir is inserted. This is done so that boundary conditions can be applied in an attempt to mimic the effect of the infinite media surrounding the reservoir. With the aim of reducing the need for discretization of large regions of the massif, a Green s functions approach was proposed for reservoir geomechanical analysis. This method is based on the use of classical analytical solutions (Kelvin s fundamental solution, Melan s fundamental solution, for example) as auxiliary solutions to solve elastically heterogeneous and nonlinear problems in fluid-saturated media. The non-linearity of the material can be due to irreversible deformations or non-linear elasticity response typical of 4D analysis. The general solution procedure relies on a discrete collocation method and an iterative fixed-point approach to build the displacement field. This method´s convergence was verified through simplified models that have analytical solutions. As the reduction in processing time is crucial for decision-makers to act in field applications, two improvements were proposed using artificial intelligence (AI) to reduce processing time of the Green s function approach. The first improvement is based on the generalization ability of artificial neural networks (ANN). Due to this characteristic, it was proposed to discretize the model with a coarse mesh of intelligent elements instead of refined mesh of traditional elements based on polynomials. The behavior of the strain field within these new elements is predicted using an ANN. To make these predictions, the neural network uses displacement boundary conditions, material properties and the geometric shape of the element as input data. The examples comparing the intelligent element approach and the traditional method were performed on a computer with 12 threads of 2,6GHz and 32GB RAM. This comparison showed reductions between five and ten times in CPU time, while maintaining the accuracy of the results. The second improvement consists in the use of auxiliary solutions that consider the heterogeneity of stratified massifs. These solutions are obtained through the training of artificial neural networks that have as output the displacement in a certain point of the stratified massif due to the application of a point load inside the massif. This ANN uses as input data elastic properties and the thickness of each layer of the massif, and of the semi-infinite media, as well as the coordinates of the point load and of the point where the displacement is to be evaluated. The use of the developed ANN-based Green’s function approach only demands the discretization of the reservoir itself, thus avoiding the discretization of other regions of the massif. Furthermore, it is possible to obtain the displacement at any point of the massif due to a pore pressure variation within the reservoir without having to solve for the other points in the massif. These two characteristics increase the efficient of the method in relation to traditional methods, such as the finite element method. A numerical example was performed on a computer with 12 threads of 2,6GHz and 32GB RAM to compare the ANN-based Green’s function approach with the traditional approach. The CPU time to obtain the solution using the ANN-based Green’s function approach was five times smaller than the that required by the traditional approach.