Reaction-diffusion coupling presents a pathway for producing nanomaterials that are precisely distributed in a reaction medium, with controlled gradients of chemistry, crystallography, and morphology. In this study, we use an automated laboratory to investigate the precipitation patterns of copper hydroxide via reaction diffusion (RD) coupling in a solution-gel system. Depending on the initial conditions, the products form continuous precipitates or oscillatory patterns typical of the “Liesegang” phenomenon. The band structures are characterized using empirical spacing metrics, which convert the complex patterns into scalar values and can thus be used for supervised machine learning in an active learning loop. The machine learning algorithm serves dual roles, providing correlation between reaction conditions and resulting precipitation patterns, which are often beyond the physics-based models, as well as dynamically evaluating the most significant areas of the parameter space. Our goal is to develop an autonomous platform wherein the user can pre-select a target product pattern, and the system converges to it with closed loop feedback. We have demonstrated a complete cycle of this process using the Liesegang precipitation of Cu(OH)2 as a test case.