Scout AI Demonstrates LLM for Battlefield Command
Scout AI has demonstrated its Fury platform, an "autonomous vehicle orchestrator" that uses a 100B+ parameter foundation model. The system allows operators to direct drones and ground vehicles in live-fire scenarios using natural language commands. The demo shows how generative AI agents are transitioning from back-office tasks to mission-critical battlefield systems.
- Scout AI emerged from stealth in April 2025 with $15 million in seed funding, co-led by Booz Allen Hamilton's venture arm, Align Ventures, and with participation from Draper Associates and Perot Jain. The company was founded in August 2024 by CEO Colby Adcock and CTO Collin Otis. - The Fury platform is described as a "Vision-Language-Action" (VLA) foundation model designed to act as an "agentic interoperability layer," allowing it to control drones and vehicles from different manufacturers by reading their documentation and generating native instructions without altering the underlying code. It is designed to operate even in environments where GPS and communications are denied. - The fiscal year 2026 defense spending bill includes $9.8 billion for autonomous and unmanned systems, part of a total IT budget of $66 billion. The Pentagon also established its first standalone budget line for autonomy, requesting $13.4 billion for AI-driven platforms. - The Department of Defense's policy on such systems is governed by DoD Directive 3000.09, "Autonomy in Weapon Systems," updated in January 2023. The policy, which aligns with the DoD's Responsible AI Strategy, requires that all systems be designed to "allow commanders and operators to exercise appropriate levels of human judgment over the use of force." - This technology fits into the Pentagon's broader efforts to accelerate AI integration, such as Project Maven, which was initiated in 2017 to use machine learning for analyzing surveillance data. The ultimate goal is to support larger command-and-control constructs like the Air Force's DAF Battle Network and the overarching Joint All-Domain Command and Control (JADC2) concept. - To accelerate adoption of commercial tech, the DoD utilizes organizations like the Defense Innovation Unit (DIU) and the Chief Digital and Artificial Intelligence Office (CDAO). In December 2024, these two bodies jointly formed an AI Rapid Capabilities Cell (AI RCC) with an initial budget of approximately $100 million to pilot generative AI tools. - The U.S. Air Force is actively experimenting with AI for command and control, releasing its first doctrine note on AI (AFDN 25-1) in April 2025. In recent experiments, AI generated battle plans 400 times faster than human staff, though some of the AI-generated plans were not viable, reinforcing the need for human oversight. - In January 2026, the Department of War issued its AI Acceleration Strategy, mandating an "AI-first warfighting force" and requiring that all contracted AI models be available for "all lawful purposes." This has created tension with some AI providers, like Anthropic, over use cases such as fully autonomous weapons without human oversight.