Here we assessed the ability of an automated sample preparation device equipped with disposable microcolumns to prepare mass-limited samples for high-sensitivity quantitative proteomics, using both label-free and isobaric labeling approaches. First, we compared peptide label-free quantification reproducibility for 1.5–150 μg of cell lysates and found that labware preconditioning was essential for reproducible quantification of <7.5 μg digest. Second, in-solution and on-column tandem mass tag (TMT) labeling protocols were compared and optimized for 1 μg of sample. Surprisingly, standard methods for in-solution and on-column labeling showed poor TMT labeling (50–85%); however, novel optimized and automated protocols restored efficient labeling to >98%. Third, compared with a single long gradient experiment, a simple robotized high-pH fractionation protocol using only 6 μg of starting material doubled the number of unique peptides and increased proteome coverage 1.43-fold. To facilitate the analysis of heterogeneous tissue samples, such as those obtained from laser capture microdissection, a modified BCA protein assay was developed that consumes and detects down to 15 ng of protein. As a proof-of-principle, the modular automated workflow was applied to 0.5 and 1 mm2 mouse kidney cortex and medulla microdissections to show the method’s potential for real-life small sample sources and to create kidney substructure-specific proteomes.