Robotic manufacturing is increasingly moving towards one-off assembly tasks. Agile manufacturing systems must be able to adapt to changing parts and designs. The addition of a soft interface, between a picking tool and a part, can improve the performance and introduce new ways of picking and manipulating parts without increasing complexity. The introduced compliance also assists when mating parts. These soft interfaces can be applied in minimalist pick and place tasks with passive gripping. This work investigates the model behind gel-based soft interfaces, and demonstrates the application for manipulating fasteners in an agile assembly task. The model predicts the limit to the holding force is from a vacuum seal, the actual holding force can be lower due to the fluid properties of the gel in confined geometries. The soft interface is shown to give adaptability for a single tool to pick multiple different sizes of parts. This interface gives success rate > 90% for well aligned cross-head bolts, and much better performance for hex keys and bolts. © 2019 IEEE.