The Agents of Things is an extension of the Internet of Things concept in which objects or things are embedded with self-reasoning intelligent software agents to provide value-added services to humans. However, a particular difficulty in designing intelligent things for the Agents of Things implementation is to determine the level of reasoning ability that corresponds to the functions of the things. To achieve this goal, we need to find an optimal match of hardware configuration with the software abilities that intelligently perform the functions of the things. In this paper, we analyze different software to identify their spectrum in optimizing self-reasoning intelligent software for the things. The result of this analysis is a software spectrum that is organized in terms of complexity and sophistication. We discuss the analytical process to arrive at the spectrum.