Scout AI

We invested in Scout AI because they're solving the autonomy problem at the layer it actually needs to be solved: the foundation model. Most robotics companies build hardware and control stacks for a single platform or a single mission. Scout is building the brain. A vision-language-action model that can be deployed across ground, air, maritime, and space systems, with natural-language coordination between them.

The defense robotics category has been constrained by the same thing that constrained civilian AI for years: specialized systems trained on narrow tasks. A VLA foundation model collapses that constraint. One model. Many platforms. Compounded operational leverage. That's the category-defining bet in defense robotics, and the teams that get it right will define the next decade of military capability.

Scout's team is unusual. The technical approach, multi-modal VLA grounded in real camera data, usable in comms-denied and GPS-denied environments, built on commercial off-the-shelf hardware, is designed for the actual deployment environment, not for a demo. The Army UxS Autonomy Contract and the Hendrick Motorsports partnership tell you the commercial and operational traction is already real.

We wrote the Series A with a long view of what defense autonomy looks like when foundation models reach deployment maturity. Scout is positioned as one of the few companies that can own that layer.

Scout AI is building the foundation model for defense robotics. A vision-language-action system called FURY that turns unmanned systems into autonomous agents across every operational domain.

FURY is a multi-domain, platform-agnostic foundation model. The same underlying model operates unmanned ground vehicles, drones, aircraft, maritime systems, and space platforms, with natural-language coordination between them. That collaborative autonomy layer, robot-to-robot communication in natural language, swarm-level coordination, and commander intent interpretation, is what makes multi-domain operations actually tractable at scale.

The technical approach is deliberately austere. Vision-based passive sensing, usually just a single RGB or thermal camera. No reliance on LIDAR, radar, GPS, or cloud connectivity. Operates in comms-denied and GPS-denied environments, which is where defense autonomy actually has to work. Built on commercial off-the-shelf hardware to enable high-rate, low-cost manufacturing pipelines. Self-improving through secure over-the-air updates and edge retraining.

The commercial trajectory is already established. Scout announced the Army UxS Autonomy Contract in August 2025 and a partnership with Hendrick Motorsports shortly after. The team is based in Sunnyvale, California, and Paso Robles, California, and is hiring aggressively across AI research, robotics engineering, and field operations.