Haptic interfaces enhance the virtual reality experience by emulating real‐world touch‐based interactions and delivering tactile feedback, while also supporting rehabilitation through interactive systems that speed up recovery. However, current haptic technologies are often built for specific uses, highlighting the need for generalized methods that allow for reconfigurable design and control. In this article, a comprehensive framework is introduced for the design and control of modular soft robotic devices. These modules can be combined in various configurations to develop a wide range of haptic devices capable of dynamically modifying their stiffness and shape to accommodate different applications, including providing haptic feedback in virtual environments and assisting individuals with hand rehabilitation needs. The design methodology of these modules is reported and two potential use cases are presented. Furthermore, it is demonstrated that how it is possible to intuitively control the modular soft robotic devices, monitor their state, and use limited sensing capability with machine learning models to enable interpretations of interactions with users and the surrounding environment.