Improved circular economy relies strongly on improved methods for recycling where automation begins to be an important factor. FlexCycle addresses the domain of soft material recycling. Here automation is largely lacking, because these materials are elastic, hard to handle, and – being “trash” – are often found in unpredictable states of degradation. This makes flexible, adaptive solutions for automation an absolute necessity. This requirement is amplified by the fact that soft materials exist in many consumer products. However, specialized recycling solutions for one product group (e.g., cloth) are usually inalterable and inflexible and cannot be used elsewhere (e.g., for cables). FlexCycle addresses this difficult problem and our main contribution is to develop flexible automation methods that allow for fast and efficient transfer between different domains of soft material recycling. We address three, structurally quite different, use cases: cloth, cables and membranes from fuel cells. The novel concept of across-domain transfer of recycling methods will be achieved by adaptive components in (1) hardware and (2) software: 1) We are developing new kinds of soft, under-actuated end-effectors and other flexible robotic tools, allowing us to handle different materials with minimal hardware-reconfiguration. 2) We are developing a generalized, novel framework for modeling soft material handling by combining a substrate-overarching structural model of soft materials with advanced graph-attention neural networks to infer the desired robotic action(s). The main impact of FlexCycle is, thus, a domain independent automation approach for recycling and reuse of soft materials allowing for reduction in system setup- and programming effort based on these novel models and hardware elements. These advancements will be complemented by a study on the impact of automation in soft material recycling concerning social, economic and environmental aspects in view of the Green Deal.