Especially for small and medium-sized enterprises, the effort for communication, planning, implementation and recommissioning is complex and time-consuming when it comes to restructuring a manufacturing plant to cope with the ever-shortening product and technology life cycles. We propose to automate the process of adapting an existing plant to the changing requirements of a plant operator by an autonomously acting machine or component. While changes so far require a well-coordinated cooperation between plant and component manufacturers and, if required, also with suppliers, the need for a change should now be detected automatically and, if applicable, this change should be carried out automatically. In this paper, subtasks for the design of autonomous adaptive machines are identified and presented within a preliminary architecture. The underlying assumption is that changes to machines and components can be supported by knowledge-based configuration technologies. These technologies, some of which are embedded in the plant or components themselves and consider the Asset Administration Shell of the German Industry 4.0 initiative, handle variability and updates by automatic derivation methods that calculate necessary changes to machines and components.