Digitalization of the building sector has become a critical topic of interest, prioritizing the decarbonization goals set for this decade. To achieve net zero emissions by 2050 according to the Paris Agreement, all new buildings and 20% of existing buildings are expected to be fully decarbonized by the end of 2030. The digital twin is an emerging concept where a physical asset and its digital copy interact in real-time, enabling various benefits such as the possibility for monitoring and analysis. Various studies have been conducted to understand how digital twins can bring value in decarbonizing buildings. In the context of evolving electricity grids, there is an evident gap in research on digital twins for grid-interactive buildings. These buildings can deliver power back into the grid and participate in the ancillary market, thus helping to avoid demand peaks, also providing flexibility. This paper conducts a systematic literature review using PRISMA guidelines on digital twins for smart grid interactive buildings, which the existing literature lacks. The study then develops a classification framework to understand the difference between a digital twin and other digital representations in the context of gridinteractive buildings based on the existing literature. It identifies and critically compares the best software tools and artificial intelligence techniques that can be used for creating a digital twin for a grid interactive building with optimal precision and accuracy in real-time. Finally, the review results are analyzed and used to develop a recommendation roadmap for developing a digital twin for a grid-interactive building.