Currently, an increasing number of residential customers have access to distributed flexible resources, including dispatchable or nondispatchable distributed generation and various flexible loads. Meanwhile, these distributed resources are also available to some local energy transactions based on internal pricing mechanism in energy communities. However, the local energy trading of distributed resources requires considerable computational capacity and professional knowledge for end-users, which makes it difficult to guarantee the trading willingness. Therefore, this article proposes an intelligent home energy management strategy for residential customers based on deep reinforcement learning techniques with consideration of internal pricing mechanism. Technically, the end-user sequential decision-making process in energy management and trading can be modeled as a Markov decision process using encapsulated resource physical status information and pricing preference information. In particular, this article considers the interactive relationship between different customers or decision-makers, capturing features of group intelligent decision that evolves as energy status change and internal price signal change. The simulation and demonstration of such an intelligent home energy management problem are provided with multiagent setup based on a quite new concept of artificial intelligence-of-things that could showcase the software and hardware implementation features at the same time. By sufficient cosimulation experimental studies, this article found that residential customers can achieve a significant improvement in their economic benefit and decision-making efficiency.