The summer counteroffensive that Ukraine launched in 2023 was, among other things, a large scale field experiment in what modern militaries call distributed autonomy. It was not, as some technophiles like to imagine, a clean demonstration of machine supremacy. Rather it was a more prosaic and instructive fusion of cheap improvisation and high-end sensors. The narrative that survives this campaign is that unmanned systems mattered because they turned information into time and time into tactical options.

At the most basic level unmanned aerial systems provided the eyes that made artillery effective. Reconnaissance drones, from simple quadcopters to longer endurance fixed wing platforms, were the force multipliers for Ukraine’s 155 mm guns and HIMARS batteries. ISR from drones contracted the sensor to shooter loop, allowing Ukrainian artillery to shift from area suppression to deliberate, target-specific strikes. That effect is both immediate and mundane. It reduces rounds expended per effect and increases the operational tempo available to a commander who can see before he commits.

But the summer campaign also featured another class of unmanned systems that will be remembered for their psychological effects as much as their physical damage. First Person View kamikaze drones, the small, high-speed “FPV” craft that Ukrainian companies and volunteer collectives mass produced, were lethal within a constrained envelope. They are hard to jam because they operate at low altitude and at high speed, and they are cheap enough to be used in quantity against exposed systems and soft-skinned logistics. Units like Escadrone and a clutch of startups produced thousands of these vehicles and trained pilots to fly them intentionally into targets. Their utility in the offensive was not that they replaced armor or aviation. Their utility was that they forced Russian defenders to allocate attention and ammunition to local air defence and to reveal positions.

At higher cost and greater risk, medium altitude systems such as the Bayraktar TB2 retained value, but that value shifted. Where TB2s had once been used to strike armor and missile batteries, by mid 2023 many of the surviving systems were being flown conservatively as reconnaissance platforms, kept out of range of layered air defences and electronic warfare suites. Their sensors and endurance still mattered. They became mobile observation posts to cue cheaper strike assets such as loitering munitions and artillery. That tactical pivot is a useful corrective to any simplistic claim that a particular platform is a war winner in and of itself. Platforms only win when they operate within margins of survivability and when the enterprise of combined arms exploits their comparative advantages.

Russian responses during the summer offensive illuminate the second-order effects of unmanned proliferation. Moscow leaned hard on Iranian Shahed loitering munitions and expanded electronic warfare measures to degrade Ukrainian command links and GPS. The result was an environment in which both sides chased resilience. Ukrainian engineers and firms raced to harden navigation, to build autonomy that could maintain a mission when communications were lost, and to develop tactics that paired “mothership” platforms with disposable strike drones. The counterpoint here is important. Adversaries will always seek the cheap asymmetric response. Ukraine’s innovation was to accept that interplay and to design systems for contested electromagnetic environments rather than for peace.

This improvisational ecology produced real operational gains but also sharp limits. Small unmanned systems have restricted range and modest payloads. They are superb for hitting soft targets, for shaping a battlefield, and for attriting logistics. They struggle against prepared defenses, complex obstacle belts, deep minefields and urban rubble. Where Ukraine sought to achieve decisive operational maneuver across tens of kilometers of prepared front, unmanned systems were necessary but not sufficient. They bought time, they reduced casualties, and they improved targeting. They did not, however, substitute for bridges, engineering units, and massed logistics. The lesson for militaries thinking about future offensives is simple. Drones can alter the character of fighting but they do not by themselves create strategic movement across terrain absent the old necessities of combined arms and supply.

There are moral and doctrinal questions that linger after the smoke clears. The rise of autonomous features and AI-assisted targeting software in Ukrainian-manufactured systems pushed the boundary between human judgement and machine execution. In many of the systems tested in 2023 the human retained the final decision to strike. Even so, the shortening of the loop between human selection and machine effect increases the moral hazard of mistakes. When thousands of small weapons blur the distinction between reconnaissance and strike, the legal and ethical frameworks that govern use of force strain at the seams. We cannot, in good conscience, celebrate efficiency without asking which institutions will hold operators and designers accountable for errors in an era of proliferated lethality.

Finally, the summer offensive was a reminder that technological advantage is transient. What counts is not the novelty of a platform but the national system that sustains it. Ukraine’s ability to translate domestic entrepreneurial energy into supply lines and training programs was decisive. It was innovation within logistics and doctrine that mattered. Future planners should note that an unmanned system is as effective as the industrial base, training pipeline and tactical doctrine that support it. To fetishize robots is to misunderstand war. To study how humans and machines extended each other’s reach in Ukraine is to learn how to contain risk and to avoid hubris.