Air Force and Space Force Ran a Two-Week AI Battle Management Experiment

In a two-week experiment, the Department of the Air Force validated a new approach to accelerating capabilities by integrating artificial intelligence and prioritizing human decision-making. Led by a U.S. Space Force Guardian, an integrated crew worked alongside Air Force battle managers and industry developers at the 805th Combat Training Squadron’s Shadow Operations Center-Nellis to test an agile acquisitions model.

“Our goal is to get the most advanced capabilities into the hands of the joint force at speed,” said U.S. Air Force Lt. Col. Corey Ellsworth, Advanced Battle Management System Cross-Functional Team integration lead. This event, dubbed the “Decision Advantage Sprint for Human-Machine Teaming” or DASH, demonstrates a focused effort to rapidly deliver validated capabilities to warfighters. It was hosted at the 805th CTS’s Shadow Operations Center-Nellis as a “whole of lab” effort, co-led by the ABMS Cross-Functional Team and the Air Force Research Laboratory.

Joint Force Integration Validated in DASH Experiment

During the two-week experiment, an integrated operational crew led by a U.S. Space Force Guardian worked alongside U.S. Air Force battle managers and industry software developers, an arrangement that Jeffrey Palumbo, AFRL C3BM Capability Area lead, described as “user-producer co-creation” that “allows for proof of concept, energizes the industrial base, and allows for early operator feedback to shape development.” Dr. Elizabeth Frost, the AFRL DASH lead, said the quantitative data and human-factors analysis supported the operators’ own conclusions, validating a core hypothesis: that the fundamental challenges of command and control are the same across every domain.

U.S. Space Force Maj. Erik Murdock, 57th Space Aggressor Squadron Operations Division chief, explained this cross-domain consistency, stating, “As a joint force, we operate as a single, cohesive team focused on the exact same objective. It isn't about the 'Air Force way' or the 'Space Force way'; it is about the 'warfighting way.'” He added that while the domains differ, the decision-making challenges are identical, and that giving a joint team common objectives and a shared set of tools made the integration happen naturally.

This finding suggests a unified approach to decision-making tools and training can streamline integration between branches, regardless of their specific areas of operation. The experiment moved beyond simulated, reactive scenarios, instead presenting participants with complex and ambiguous problems requiring proactive decision-making under pressure. “We are moving beyond simple, reactive scenarios and presenting our warfighters with complex, ambiguous problems where they must be proactive, weigh trade-offs, and make decisions when there is no perfect answer,” said U.S. Air Force Col. John Ohlund, ABMS Cross-Functional Team director. “This approach is essential for developing the cognitive superiority our joint force needs to out-think and outmaneuver a peer adversary in a high-end conflict.”

Ohlund framed that shift as the point of the whole series. The future of battle management, he said, “isn't just about accelerating our reaction time; it's about fundamentally changing how we make decisions,” and the experimentation, which focused on game theory, is set to continue in future sprints. That sets the standard the sprint had to meet: not a faster button-press, but a better call under conditions where no option is clean. It also puts the burden on tools that can absorb the data handling without taking the decision away from the person accountable for it.

The transition path off the unclassified floor was designed in from the start. “This entire framework was built from the ground up to ensure that once a capability is validated here, we can securely and efficiently transition it to the classified networks where our Airmen and Guardians will actually use it in a high-end fight,” Ellsworth said. Software developed by AFRL’s Information and Spectrum Warfare Directorate assessed each experimental run in real time, both with and without vendor solutions in the loop. “This real-time assessment allowed vendors and operators to see process execution metrics associated with each trial to give immediate feedback on the benefits of introducing technology solutions into battle manager’s decision-making process,” said Ryan Hilliard, AFRL Information and Spectrum Warfare Directorate C2 Research Area Lead.

AFRL “Orchestrator” Enables Agile C2 Architecture

The experiment’s technical foundation rested on a government-owned software interface, an AFRL-developed “orchestrator” that let multiple competing industry applications be tested inside a single modular ecosystem. “Successfully demonstrating this 'plug-and-play' ecosystem was a critical milestone,” said Kevin McFarland, AFRL Advanced Capabilities Directorate project management lead. “Government ownership of the orchestrator significantly decreases integration bottlenecks. We are no longer locked into a single vendor's closed architecture.” Because the environment itself is unclassified, vendors could iterate alongside operators on the floor while the architecture was built from the outset for a secure hand-off to the classified networks where the tools would actually be used.

Realism was the other half of the design. “This isn’t a scenario from a video game; it is one crafted to challenge operators and engineers alike,” said U.S. Air Force Lt. Col. Wesley Schultz, 805th CTS commander. “We reflect the real-world challenges of geography and logistics, and even real-time changes in the rules of engagement, because that reflects the reality of modern warfare.” The scenarios were written by the ShOC-N’s own team of veteran operators, drawing on more than 50 years of collective experience in tactical and operational command and control.

Human-Machine Teaming Reduces Operator Cognitive Load

The clearest operator-level result was about attention rather than speed. Because the sprint was run as a “whole of lab” effort, subject matter experts from several AFRL directorates and the 711th Human Performance Wing sat on the operations floor analyzing operator workload data as the runs unfolded, rather than reconstructing it afterward. That arrangement was intended to let the services scientifically evaluate and optimize the cognitive efficiency of human-machine teaming before any of it reaches a warfighter. The tooling under test was the kind that automates target action recommendations and generates rapid courses of action — work a battle manager would otherwise assemble by hand while the clock runs.

“Both Airmen and Guardians are tasked with synthesizing vast amounts of complex data and prioritizing actions under intense time pressure,” Frost said. “This event demonstrated that a well-designed human-machine team can effectively free up an operator's cognitive resources.” Those freed resources are the point, letting an operator spend judgment on the decisions that have no clean answer instead of on assembling the picture. Schultz put the change in career terms, calling the sprints “the biggest shift in battle management in my career” and describing the approach as taking what technology allows and shaping it to magnify the skills and training of warfighters.