Faster than the threat adapts
At the AUSA annual meeting on October 13, Army Col. Michael Penn described Task Force Easy, a new Pentagon unit built to iterate on weaponized small drones and small uncrewed ground vehicles across the military. His assessment of the current state was blunt. "We're slower, less effective. We don't outpace our enemy and our threat in how quickly we adapt new technology." The goal he set is what he called adaptation dominance. Buying 1,000 drones is the easy part. The task force is meant to build an institutional and industrial engine that learns, updates, and fields better capability faster.
The task force starts with first-person-view and one-way attack drones. It leads the pacesetter work on the Chief Digital and AI Office's Project Swarm Forge, which aims to deliver collaborative autonomy at the edge. It is also standing up a drone team for training, tactics, techniques and procedures, and standards, and that team will support the combat training centers when they evaluate new equipment.
Volume is no longer the constraint
The buying side is moving. Phase III of the Drone Dominance Program plans to purchase about 60,000 one-way attack drones for $450 million, with an option for up to 40,000 more, at fixed prices of $5,000 for a deep strike drone and $3,000 for a close-quarters drone. Vendors have to show they can build at least 8,000 in five months. In September the Army's Drone Dominance director said of equipping squads, "We do have the volume to do that." The work that remains, he said, is "getting everybody trained at the appropriate level." An Army spokesperson added that the drones have not yet reached every squad.
The Phase III solicitation also shows where the program wants to go next. It asks for drones that work where satellite navigation or communications are disrupted, automated target recognition, and one operator controlling multiple aircraft. Standardized interfaces are preferred but not required. Swarm Forge goes further. Its solicitation defines the goal as multi-vendor command, control, and autonomy with "minimal operator intervention," kept under "meaningful human command."
An engine that learns needs something to learn from
Adaptation is a loop. A unit flies a sortie, something works or fails, the lesson reaches the people who can change the drone, the tactic, or the training, and the change comes back to the unit. The loop is only as fast as its slowest step, and the slowest step is usually the first one: getting an accurate account of what happened out of the field.
Today most of that account lives in the operator's head, a notebook, and whatever the vendor's ground station happened to log. Which target was assigned, and by whom. What the link looked like when the video froze. Whether the second drone was sent because the first one missed or because nobody knew the first one hit. A squad that flies drones from three vendors has three partial logs and no shared one. A standards team can write excellent procedures and still be working from anecdotes.
What the squad layer records
EdgeLance treats each sortie as a task with a record. A target or a reconnaissance task is offered to the assets that can take it. Assets bid based on position, payload, and endurance. The winner holds the task on a lease, and the result comes back as a receipt on that task. Every asset sits on one roster with its battery and health. The record of who tasked what, which asset accepted, and what came back is written on the team's own devices and moves over the team's mesh, so it exists whether or not the squad had a link to anyone else that day. The assignment logic was built and stress-tested internally against counter-swarm scenarios.
That is also what one operator controlling several aircraft requires. The operator stops flying sticks and starts approving tasks, which only works if the software knows which aircraft holds which task and takes it back when an aircraft goes quiet. The autonomy software being tested in efforts like Swarm Forge governs how aircraft fly and cooperate. The tasking layer sits above it and answers a different question: which task, whose task, and what happened. The two are complementary. Better autonomy makes each task more likely to succeed, and a shared task record lets a mixed fleet from several vendors be run as one group.
Write the record into the standard
Task Force Easy's drone team will define standards for training and employment. One of them should cover what a squad records for every sortie and in what form, so that the lesson from a failed strike in one brigade reaches the vendor and the schoolhouse in days. The requirement is modest: task, asset, time, link conditions, outcome, and who approved it, kept on the unit's own devices and released upward by the unit.
The drones will keep arriving. How fast the force adapts depends on how much of each sortie survives as data. The jobs around the airframe are laid out in The drone arrived before the pilot and in the whitepaper Running an Attritable Force. See how tasking and the mission record fit together on the platform page, or browse the capabilities.