Manned-Unmanned Teaming

Manned-unmanned teaming (MUT) is an operational concept in which crewed aircraft and autonomous drones fly coordinated missions together, with AI enabling the unmanned platforms to act as force multipliers for their human-piloted counterparts. The concept represents one of the most consequential applications of agentic AI in defense: a single pilot directing multiple autonomous "loyal wingmen" that can scout ahead, jam enemy radar, carry weapons, or absorb threats — all at a fraction of the cost of additional crewed fighters.

The "loyal wingman" concept sits at the core of MUT. These are jet-powered, semi-autonomous combat drones designed to fly alongside fourth- and fifth-generation fighters. Unlike traditional remotely piloted aircraft (like the MQ-9 Reaper), loyal wingmen operate with significant onboard AI, making real-time navigation and tactical decisions with minimal human input. The human pilot functions as a mission commander rather than a direct controller, issuing high-level objectives while the AI handles execution — a pattern directly mirroring the multi-agent systems paradigm in civilian AI.

Programs are racing forward globally. Australia's Boeing MQ-28 Ghost Bat completed over 100 test flights by early 2025, with air-to-air missile launch tests planned for 2025–2026. The U.S. Air Force's Collaborative Combat Aircraft (CCA) program, informed by years of experimentation with the Kratos XQ-58A Valkyrie (estimated at just $2–4 million per unit versus $80+ million for an F-35), aims to field autonomous wingmen at scale. The Valkyrie has demonstrated coordinated swarm operations under the PAACK-P program. Turkey's Bayraktar Kızılelma entered mass production in late 2024, targeting 10 units by 2026. India's HAL CATS Warrior completed its first engine ground run in January 2025. Europe's Airbus Wingman is targeting the 2030s as part of the Future Combat Air System (FCAS). China's Feihong FH-97 and Russia's Sukhoi S-70 Okhotnik-B round out a competitive global field.

The AI challenge in MUT is uniquely demanding. These systems must process sensor data, maintain formation, avoid threats, and execute tactical maneuvers — all in environments where communication links may be jammed or degraded. This requires robust onboard inference capabilities, not cloud-dependent AI. The drones must handle the equivalent of embodied AI in the most adversarial conditions imaginable: supersonic speeds, electronic warfare, and dynamic multi-threat environments. Failure modes are lethal — as demonstrated in October 2024, when Russia deliberately shot down its own S-70 Okhotnik-B after losing control of it over Ukraine.

MUT changes the economics of air power. If a loyal wingman costs $3 million and an F-35 costs $80 million, a fleet of 10 drones plus one crewed fighter delivers far more combat capability per dollar than two crewed fighters. The U.S. Replicator initiative explicitly aims to field large numbers of autonomous systems to offset adversaries' numerical advantages. The drones are designed as "attritable" — valuable enough to recover, cheap enough to lose. This calculus fundamentally reshapes procurement, training, and doctrine, and connects directly to broader debates about autonomous weapons and the role of human judgment in lethal decisions.

The trajectory is clear: within a decade, no major air force will operate without AI-enabled wingmen. The technology draws on advances in reinforcement learning, swarm intelligence, and computer vision, applied under constraints that make most civilian AI applications look simple by comparison.