We aim to conduct independent research and deliver model and data agnostic tools to improve the performance of your neural networks. We believe that the current common understanding of the inner workings of neural networks is rather limited. Most people believe that neural networks work because “the math say they do”. While we are not ones to argue with math, we approach the understanding of neural network operation from a geometric perspective. This can provide us with insight regarding WHY and HOW various neural network training techniques work and improve upon them. We already provide improved alternative neural network layers and we plan to provide in-between layers, improved optimizers, improved error functions and more. We aspire to fund our advancements through a “Research As a Service” model, by offering early access to our research results through our unique premium licenses, which do not cost more, but require a longer commitment.