The manufacturing of medicines is often expensive because only a few successful products reach the market and have to compensate the costs of the drugs that failed during development. Reasons for failure are products that cannot be produced cost-effectively in process scale, which often turns out late during development, i.e. during clinical phase trials. Therefore, we are investigating within the Fraunhofer Attract group "FAST-PEP" how predictive models, e.g. of chromatography, can be used to render process development more successful even in early stages and avoid high costs of late stage failures. The predictive models are being applied to individual product candidates, e.g. vaccines, to compare different purification strategies in silico and verify the most promising ones by high throughput screening.