Aggregation of the β-amyloid (Aβ) peptide into toxic oligomers is considered the primary event in the pathogenesis of Alzheimer's disease. Previously generated peptides and mimetics designed to bind to amyloid fibrils have encountered problems in solubility, protease susceptibility and the population of small soluble toxic oligomers. We present a new method that opens the possibility of deriving new amyloid inhibitors. The intracellular protein-fragment complementation assay (PCA) approach uses a semi-rational design approach to generate peptides capable of binding to Aβ. Peptide libraries are based on Aβ regions responsible for instigating amyloidosis, with screening and selection occurring entirely inside Escherichia coli. Successfully selected peptides must therefore bind Aβ and recombine an essential enzyme while permitting bacterial cell survival. No assumptions are made regarding the mechanism of action for selected binders. Biophysical characterisation demonstrates that binding induces a noticeable reduction in amyloid. Therefore, this amyloid-PCA approach may offer a new pathway for the design of effective inhibitors against the formation of amyloid in general.