A key problem that we are facing in cosmology nowadays is that we cannot make accurate predictions with our current theoretical models. We have all of the pieces of the standard model but it doesn't have an analytical solution. The only way to have accurate predictions is to run a cosmological simulation. Then, why not use these simulations as the theory model? Well, for one main reason, if we want to explore the full parameter space comprised in the standard model, we need thousands of such simulations, and they are terribly computationally expensive. We wouldn't be able to do it in years! In this talk, I will tell you how in the last few years we have come up with a way to circumvent this problem.