This paper introduces PyWiSim, a Python-based simulation framework designed for wireless systems that falls somewhere between a link simulator and a system simulator. Link simulators model all communication layers in detail, making large-scale simulations computationally expensive. On the other hand, system simulators typically perform throughput calculations for a given simulation scenario, allowing simulations with many devices but providing little detailed information. With this compromise between these two classes of simulators, PyWiSim seeks a simulator that enables simulations with a large number of devices but modeling the most relevant aspects of the system with a certain level of detail. This framework is well-documented and allows for the easy addition of new wireless channel models, traffic generators, scheduling algorithms, etc. Being built in Python—a language widely used in artificial intelligence (AI) applications—PyWiSim facilitates the natural integration of AIbased algorithms into wireless simulations. To demonstrate this versatility, we present an example of a scheduler developed using deep reinforcement learning, specifically the Deep Q-Network (DQN) algorithm. It natively supports multislice, a fundamental feature of modern networks like 5G, and provides a flexible architecture that allows extensions to various wireless technologies, as demonstrated in this paper. Finally, we also present some graphical results obtained from PyWiSim to illustrate its capabilities.