As Operational Medicine (both civilian and military) must be delivered in areas of operations that are larger and more difficult to access than existing medical infrastructure, capacities become fragile and the necessary training to manage life-threatening emergencies more complex. Less experienced first-responders are increasingly expected to manage complex patients for longer periods of time, as evacuation to definitive care is extensive, impeded, or both. One potential approach to address this capacity problem is to bring the expertise to the field, if not physically, at least remotely. Thus, telemedicine, and more specifically in telementoring where an expert surgeon is paired with a medic, allows a less-experienced provider to be remotely guided through the necessary steps to perform life-saving interventions (LSIs) if absolutely necessary. Realistically however, many forward operating bases will lack the required bandwidth to enable such a paradigm. To address these limitations, two categories of artificial intelligent (AI) agents have been proposed: personal assistants or copilots (i.e., Alexa, Siri, Watson) that are strong reasoners when all information is provided in the form of natural and written language and AR/VR AI trainers that can interact more naturally, and achieve broader awareness of the user and environment (vision, sound, gestures). However, key gaps in implementing these agents are their poor or absent modeling of the user and lack of understanding of how the user interacts with contextual information, lacking ability to change and be changed by the environment. To address these gaps, personal surrogate agents have been proposed as a third approach, such as the Trauma THOMPSON System. Referred to as Copilots Surgical Mentors (CoPiSMs), these egocentric systems of sensemaking and instruction—reasoning within and between domains (e.g., LSIs) are capable of communicating and adjusting decisions at the human level. CoPiSMs systems can be both generalist and/or specialist according to the user’s state and need. These AI mentors are agents with a defined set of skills (e.g., airway management, ship maintenance) that are also capable of learning new skills or converting old skills to new domains. This paper reviews the state of the art of such new technologies, such as AI copilots and assistants with augmented reality (AR) for telemedicine. The focus of the review is placed in context of first-responder capability enhancement in challenging environments. In any austere or resource-constrained deployment, these systems hold the promise to provide remote mentoring and automation, to extend, multiply, and “transfer” physician expertise to responders closer to the point of injury or critical illness through step-by-step guidance through LSIs for patient real-time treatment.