Simulation & Analysis with Generative Entities (SAGE)

Stottler Henke proposes to develop the Simulation & Analysis with Generative Entities (SAGE) system, applying generative AI methods to drive realistic, tactically sound, adaptive threat behavior during NGTS scenario runs and to analyze blue communications during scenario runs for performance assessment immediately after. The proposed technology will enhance training and readiness through two key improvements. First, more adaptive, realistic threat behaviors that are easy to direct in real-time will result in more efficient use of the instructor’s time, less time required of NGTS support staff, more effective training for students, and an improved capacity for NGTS to be applied to mission rehearsal both ashore and afloat. Second, automatically evaluating friendly comms during a scenario provides feedback that directly develops communication skills in trainees, resulting in improved situational awareness and decision-making across the board. The proposed SAGE system relies on agentic workflows to build a collection of LLM-based agents that accomplish both of these tasks within a shared architecture integrated with NGTS. Proof of concept development in Phase I is facilitated by our team’s combination of subject matter expertise and experience implementing a wide range of AI applications, natural language analytics, and training debrief tools.