In an attempt to counter growing cybersecurity challenges, there is an increasing need for inculcating IoT devices in Small and Medium-sized Enterprises (SME) to improve efficiency by streamlining operations. While traditional mechanisms have been fairly effective, they are not practically feasible to be applied everywhere because they are expensive, rely on a centralized infrastructure to work, and can’t update automatically to deal with developing technology that can prove to be a threat. This means that existing IoT networks are vulnerable to advanced cyberattacks which can prove to be detrimental to the data and operations of an SME. This paper is an attempt to address these challenges, which focuses on proposing an AI-powered cybersecurity framework tailor-made for SMEs using IoT devices. The framework is a mixture of edge computing, Zero Trust Architecture (ZTA) and lightweight machine learning models to bring about a real-time threat detection and mitigation mechanism. This involves local data processing on IoT devices, thereby reducing latency. A continuous authentication mechanism is also incorporated which makes the solution cost-effective and scalable while simultaneously improving security. While this approach focuses on increasing the security of existing IoT networks, it ensures that there is no requirement of extensive resources for providing this solution, with the key goals being sustainability and ability to handle future issues.