The military is exploring the use of new technologies to improve the training of a large pool of personnel to maintain and repair a variety of complex equipment. Achieving technology integration in maintenance training requires developers to clear three human performance hurdles: physical usability, learning content efficacy, and a path to integration with existing learning methods. Augmented reality technologies smoothly project explanatory visuals (e.g., text, directional arrows, videos, and 2D and 3D animations) over the workspace. Interactive technologies verbally dictate steps to the user and respond to users’ spoken commands (e.g., “Computer, repeat step.�). In developing AR-Mentor, an innovative maintenance training technology that combines augmented reality and interactive dialogue technologies, a team of engineers and education researchers encountered and responded to each of the human performance hurdles. In this paper, we show how these human performance hurdles were addressed and how they informed the refinement of the AR-Mentor in two rounds of system development and testing. The AR-Mentor system provides a Heads-up and Hands-free experience to permit a user to train with real equipment. The AR-Mentor system consists of a compact computer, head worn cameras, microphone, ear-buds, and augmented reality eyewear. The learning content addressed in the two rounds of testing focused on both basic training in maintenance procedures and more advanced training in troubleshooting. The performance evaluation measured usability, time to learn, and the relative learning achievement in procedural knowledge and troubleshooting reasoning between business-as-usual instruction and technology-assisted learning conditions. The paper concludes by presenting key human performance concepts for trainers and vendors of complex equipment systems to consider when designing technological content for presentation with newer automated instructional technologies such as AR-Mentor.