Award

Large language models (LLMs) and natural language processing are rapidly evolving and expanding fields. Tools developed utilizing LLMs are increasingly effective across multiple domains. The use of a network of LLM-based agents to improve in-theater communications for operators, pilots, and autonomous vehicles is an important emerging area of investigation.ĀĀOperator-like agents using LLMs and natural language processing to perform functionally defined roles will allow the development of a low-cost, realistic, and configuration-controlled training environment for in-theater communications. This capability will make possible the instruction of autonomous vehicle operations over secure voice or natural language text-based communications, improving integration between crewed and autonomous systems.ĀFor Phase I, ExoAnalytic's ChatTTP will provide an end-to-end framework that allows operators the ability to instruct networks of agents to perform various communications and TTPs for theater operations.ĀThis framework will provide a base agent for each operator position defined for theater operations trained on their specific functions using documents related to the specific role. The Phase I delivery will be a prompt-based graphical user interface (GUI) which provides communication logs between agents based on the scenarios given and a containerized package of agents.