Printing complex and unique form have an increasing demand in industry, corresponding to this, metal 3D printing is a good alternative. It offers a possibly to print pieces without having to create a new production chain and facilitates the access to them. However, as a growing demand and a relatively new method, there is a room for improvements. This thesis aims to answer this request, researching how to improve the efficiency through heat simula-tion and AI. Before presenting the results, here are the major steps to achieve that. Firstly, creating a heat simulation using a voxel representation and a heat equation to fit each piece geometry. Using Robot studio to simulate the printing, everything is set up to train an AI except data. To counter this problem, the use of reinforcement learning, which doesn’t re-quire much data to be trained, will be the model of AI. The results of this thesis work are an improvement in time saving compared to a simple algorithm, at best 13%-time savings, as well as an accurate representation of pieces and an acceptable thermal simulation