Softbots, or virtual assistants, represent a disruptive technology that can enhance productivity and efficiency in many industries. This digital transformation can increase the maintenance aspects of capital goods, offering streamlined solutions for troubleshooting in diverse platforms and industries. In this regard, the focus of this article is to explore how softbots can be developed with cognitive computing, allowing them to operate across different platforms, such as VR, AR, mobile, or web. The literature provides models to provide interoperability to softbot implementation. By conducting a use case application in the food industry, this research validates the architecture proposition considering a machine for processing animal protein, as well as suggests three scenarios related to cognitive computing: reactive, planned and proactive modes. The findings of this research can help business improve their productivity and streamline their troubleshooting process through the implementation of an interoperable softbot that is capable of operating on different platforms.