A simple assembly system consists of two subassembly cells. One unit p
roduced from each cell is required to assemble a product. The producti
on processes at the cells are modeled as point processes with controll
able stochastic intensities. We identify a threshold control and estab
lish its optimally using an approach that incorporates dynamic program
ming arguments into stochastic intensity control. With combined power
of both, the optimality of the threshold control is established in a p
oint process setting that is much more general than the Markovian sett
ing usually assumed in dynamic programming.