We present a computer-assisted approach to approximating coarse optimal switching policies for systems described by microscopic/stochastic evolution rules. The "coarse timestepper" constitutes a bridge between the underlying kinetic Monte Carlo simulation and traditional, continuum numerical optimization techniques formulated in discrete time. The approach is illustrated through a simplified kinetic Monte Carlo simulation of NO reduction on a Pt catalyst: a switch between two coexisting stable steady states is implemented by minimal manipulation of a system parameter.