KSX can distribute its computational load over a local area network, or an internet virtual private network. This allows the user to configure as many neural networks and optimizers as desired and run them in parallel in real time and on-line. Each neural network and optimizer can be assigned to a client machine when it is started, or KSX can determine the client with the least CPU load and distribute the task to that machine.