We show that PXRD data acquired by robots can be matched with crystal structure prediction data, but even for simple compounds (e.g., benzimidazole), the algorithm can identify the wrong structure (table). Suggests we need improvements to sample prep & algorithm Chemical Science