A simple model of the organization of atmospheric moist convection by cold outflows is presented. The model consists of two layers: a lower layer where instability gradually builds up, and an upper layer where instability is rapidly released. Its formulation is inspired by Abelian sandpile models: instability and convection are both represented in terms of particles that are coupled to a lattice grid. An excess of particles in the lower layer triggers a particle release into the upper (cloud) layer. Particles in the upper layer also induce particle movement in the lower layer: this reverse coupling represents the effect of precipitation and the associated cold outflows. The model shows two behavioral regimes. Activity is scattered when the reverse coupling is weak, but when it is strong, convection forms cellular patterns. Though this model does not contain a detailed representation of physical processes in convection, it captures some key dynamical features of precipitating convection seen in satellite observations and LES studies. These include the formation of open cells, temporal oscillations in convective intensity, hysteresis, and the effect of precipitation on the scale of convection. We argue that an object-based representation of convection may be able to capture properties of convective organization that are missing in traditional parameterizations.