This paper presents the progress of the PREDIVIS project, on the development of a novel energy disaggregation hardware/software tool towards energy efficiency, anomaly detection and predictive maintenance. The PREDIVIS project involves the use of an edge-cloud hybrid computing architecture, hardware accelerated algorithms, machine/deep learning, and big data approaches for the analysis of high frequency electrical loads. The project will deliver an advanced and innovative solution for energy disaggregation intended for industrial applications, commercial buildings and households. A main scope of the project is to address the problem of predictive maintenance by providing a cost-efficient solution with industry 4.0 features.