A kill chain measured in seconds cannot depend on processes measured in weeks or months. The Pentagon recognizes that drone proliferation and AI-enabled lethal targeting are reducing decision cycles and increasing the premium on speed. The 2025 drone and acquisition reforms and the 2026 Artificial Intelligence Strategy call for faster procurement, fewer integration barriers, open architectures, continuous experimentation, and greater access to emerging technology.

Yet the experience of myself and fellow soldiers in the 1st Brigade Combat Team, 82nd Airborne Division (1/82), before and during its deployment in support of Operation Epic Fury shows that this reform agenda falls short at the tactical edge. For all the progress made in the Pentagon and combatant commands to speed up the kill chain, those at the very tip of the spear still face delays in integrating drones, targeting software, networks, and fires systems. Policies designed to accelerate innovation have not consistently given brigade-level formations the authorities, technical support, and development access needed to assemble and adapt integrated kill chains during operations. The Army has the technology to compete, but institutional latency constrains tactical formations’ ability to adapt.

Tactical commanders need the authority, access, and technical support to control and adapt their own kill chains, while maintaining targeting authorities and human judgment over the use of force. Closing this gap requires delegating purchasing authority to brigade commanders, delegating integration authority to division commanders, providing dedicated technical support to brigades, and giving tactical formations direct access to AI development.
 

Sign Up for Our Newsletter

What We Learned Building a Digital Kill Chain

A kill chain is the process of detecting and confirming an enemy target, transmitting the information to a shooter, engaging the target, and assessing the effects. A fully digital kill chain would network sensors and shooters within a shared computing environment, enabling AI to rapidly recognize targets and autonomously pair them to appropriate firing systems, while keeping trained operators responsible for lethal decisions. This ideal is still far from the reality experienced by soldiers today.

For example, 1/82’s legacy kill chain process involved a sensor, such as a drone, using full-motion video to observe possible targets. Junior paratroopers, with limited training and less than a year of experience in drone-enabled kill chains, watched the video, wrote down the grids in their notebooks, and verbally transmitted them as a separate fire mission through multiple physical command posts. Every manual handoff introduced delays into the process, forcing capable and motivated paratroopers to spend precious time acting as data couriers and analysts.

Possible workarounds like quick-fire channels, introduced during our integration and testing efforts, shortened the kill chain by bypassing intermediate command posts to send missions directly to firing units. However, the underlying process remained unchanged as these nets still relied on voice communications, manual data transfers, and analog systems. Additionally, removing intermediate steps concentrated more traffic at the receiving end, increasing the burden on personnel processing fire missions and network congestion.

We found sensors, cloud services, targeting software, and fire direction systems were developed and fielded for different purposes, on different timelines, and with different technical interfaces. Connecting them required bespoke network configurations and data links, constant troubleshooting, and repeated testing before data could flow reliably and cleanly from one system to another. But despite investing approximately 3,500 paratrooper man-hours in integration, testing, and experimentation, and not on invention, 1/82 achieved only a partially digitalized kill chain. Drone video and targeting data could be routed to an IL5 cloud environment, but operators still had to check and correct data because of integration and translation problems. They also had to manually transmit targeting packets to lethal fire direction systems because systems could not reliably exchange information. These technical workarounds left the process too slow for combat operations in Central Command.

1/82 also had access to Palantir’s Maven and successfully integrated some, but not most, of its sensors and digital systems. However, even when we established links, data cleanliness and translation problems persisted. For this integration work, the lack of direct on-site engineering support made it slow and labor-intensive to troubleshoot those links or establish additional ones.

To be clear, 1/82 is only one formation operating with its unique mix of systems, personnel, and operational demands, so its experience is not evidence that every tactical formation encounters identical barriers. For example, the 4th and 25th Infantry Divisions have worked with dozens of industry partners and multiple Army program offices during heavily supported prototyping and experimentation efforts, with the 25th demonstrating a fully digitized fires chain. But those efforts benefited from purposeful funding, direct and robust technical support, and access to development partners that tactical formations do not routinely possess. They also occurred largely in training and experimentation environments, where integration can be bounded in ways that operational adaptation cannot.

Department Reform to Tactical Implementation

Rapid and successful technology adaptation depends in part on faster contracting for technical services and necessary software and equipment, easier system integration, open architectures, and continuous experimentation. For 1/82, connecting components that worked independently required continuous data engineering, network reconfigurations, and testing. Contracted engineers could help with that work, but hiring them would not, by itself, resolve delays in obtaining integration approvals.

More broadly, the challenge faced by 1/82 was translating that strategic intent into authority and support at the brigade level. The Army command controlling the funds restricted their release for brigade priorities, integration approvals rested with different authorities depending on the systems involved, and technical support was only intermittently available. The consequence was a gap between enterprise-level reforms and the capabilities available to tactical formations in combat.

Many of the enterprise-level controls serve legitimate purposes. Acquisition oversight protects public funds and limits dependence on individual vendors; cybersecurity and system-approval processes protect our networks; and targeting safeguards reduce the risk of error as data moves toward lethal effects. The challenge is balancing those assurances against the operational risk created when adaptation and approval timelines lag behind battlefield requirements.

The following four recommendations would help tactical formations rapidly adapt and improve their kill chains while preserving necessary safeguards.

Delegate Financial Authority to Tactical Commanders

The July 2025 drone reforms delegated greater authority over small drones to lower echelons and emphasized faster modification by frontline personnel. Our experience suggests that the same approach should extend to the commercial software and services required to connect those systems into a functioning kill chain.

For 1/82, the immediate constraint was delegated authority to use available funds. Curtailed financial authorities limited the brigade’s ability to contract for services and purchase commercially available equipment and software. Even modest purchases of AI-enabled software systems and networking tools require procurement timelines measured in fiscal quarters, delaying the integration of existing technologies and preventing tactical commanders from adapting kill chains at the pace of combat.

To align financial authority with battlefield realities, brigade commanders need funds made available for their integration priorities and authority to approve how those funds are used. This should include individual purchases of commercially available equipment and software up to $50,000 and individual technical service contracts up to $100,000, very reasonable initial limits. That delegation would remain bound by existing fiscal controls, competition requirements, cybersecurity standards, and applicable safety and fielding requirements. Where practical, purchases should also include open interfaces and government data rights to reduce dependence on a single vendor and preserve future integration options. This authority would help close time-sensitive tactical integration gaps, while enterprise acquisition remains responsible for enduring programs and common architectures. Larger software requirements and sustained engineering support would require separate funding and approval arrangements.

Delegate Integration Authority to Operational Commanders

The January 2026 Artificial Intelligence Strategy and other initiatives referred to in this article call for fewer integration barriers. For 1/82, connecting tactical data environments, fires systems, and external data services still required multiple high-level approvals. Even minor data-routing adjustments required weeks of coordination. In some cases, 1/82 had to coordinate with department-level offices to resolve tactical integration issues, adding distance and delay to problems arising in an active combat zone.

That delay creates operational risk, but the legacy approval process also addresses legitimate concerns. Connecting previously separate systems can introduce cybersecurity vulnerabilities or degrade data integrity. Those risks become especially consequential when digital workflows feed targeting and weapons employment. Any faster approval process should therefore preserve technical review and established targeting safeguards.

Therefore, the first division-level commander in the operational chain should receive delegated authority, within established cybersecurity and risk-management standards, to approve tactical system connections and conditional Authority to Operate modifications. Previous efforts to improve the authorization process largely focused on accelerating or streamlining it at the enterprise level. This proposal takes a different approach by moving bounded authority for time-sensitive tactical integrations closer to the operational problem.

A general officer headquarters can bring together operations, intelligence, fires, cybersecurity, legal, communications, and technical expertise to assess both the risks of a proposed connection and the operational consequences of delaying it. Under this delegated authority, approvals would require post-implementation technical review and be scoped and time-limited, with defined interfaces, monitoring requirements, rollback procedures, and mandatory follow-on review. This approach allows commanders to accept operational risk without turning temporary adaptations into permanent exceptions to cybersecurity standards. Delegated authority enables bounded modifications to be assessed and approved in days rather than weeks or months, while preserving oversight and aligning integration authority with operational responsibility.

Push Technical Expertise to the Tactical Edge

Tactical integration requires both delegated authority and dedicated technical expertise. The Pentagon’s November 2025 Acquisition Transformation Strategy distinguishes between technology readiness and integration readiness and acknowledges that it lacks a fully resourced pathway for quickly integrating commercial technologies into integrated kill chains. Our brigade experienced that gap directly. A significant share of the roughly 3,500 paratrooper hours invested before deployment went to system integration and data-pathway troubleshooting performed by soldiers not doing their normal operational duties. In some cases, integrations had to be repeated across hundreds of platforms rather than applied across an entire fleet, an incredibly labor-intensive process. Specialized engineering support was intermittent and often difficult to access, forcing us to rely on informal relationships that sometimes left paratroopers waiting days or weeks to resume testing.

Faster purchasing and approval address only part of the integration problem. Creating interfaces, translating data, configuring networks, and troubleshooting failed connections still require engineers who can work alongside operators. Dedicated direct-support technical teams should therefore support brigades before and during combat operations. They don’t need to be organic to the brigade, but they should commit personnel and time to direct support of the brigade commander’s integration priorities. In the near term, these teams would rely on contracted engineers.

Over time, the Army should build a military and government civilian technical workforce with requisite system access and rights. They would engineer data links, automate targeting workflows, troubleshoot integrations, and continuously adapt digital kill chains as systems and operational requirements change. Embedded engineers would also strengthen assurance and reduce risk by identifying unsecure configurations and risky integrations before commanders accept them operationally. Persistent technical support would make continuous integration and adaptation a routine brigade function.

Give Tactical Formations Direct Access to AI Development

Tactical formations should be central participants in developing emerging AI capabilities. Brigades can identify operational needs, contribute sensor data, and test capabilities with developers responsible for the underlying software and models. Currently, their limited access to these developers restricts opportunities to use operational data to refine target recognition models or test AI-assisted fire direction workflows, for example. They also struggle to access dependable warfighting AI tools on “sensitive but unclassified” networks, along with the experimentation programs and events that shape them. This disconnect slows the feedback loop between battlefield experience and AI development and leaves tactical formations dependent on software that may not reflect operationally relevant targets or environments.

Programs like Swarm Forge and Agent Network establish an approach that pairs operational experimentation with rapid AI development, but it does not yet reach tactical formations consistently outside discrete experimentation efforts. Brigades should therefore have a formal role as continuous development partners, with regular and direct access to supporting developers. Participation should focus on testing and refining warfighting AI programs on approved sensitive but unclassified networks and providing operational data and feedback. If existing equipment cannot run software processes, units should be given high-bandwidth connections to cloud computing resources or more capable computing equipment within their formations. Access would allow operational data generated in Central Command to inform model refinement and workflow testing, helping AI capabilities evolve on timelines relevant to battlefield adaptation.

Conclusion

The Pentagon’s recent drone, acquisition, and AI reforms are movement in the right direction. However, 1/82’s experience shows that success will ultimately depend on whether those reforms change conditions at the tactical edge, where formations should continuously integrate quickly changing technologies that do not arrive fully interoperable. Success also requires moving risk decisions to echelons that can weigh warranted safeguards against the operational consequences of delay. That requires bounded purchasing authority, integration decisions close enough to operations to matter, dedicated technical expertise, and routine participation in AI development.

Ultimately, the measure of the Pentagon’s reforms should be whether a brigade can adapt its kill chain at the speed of the battlefield, a metric at which we are falling short.

Write for Cogs of War

Lucas Ziller is a U.S. Army major and infantry officer with experience in conventional and special operations assignments. Most recently, he served with the 1st Brigade Combat Team, 82nd Airborne Division, in U.S. Central Command, where he helped integrate AI-enabled systems and autonomous technologies into operations. He is currently a Ph.D. student in Public Administration at North Carolina State University.

Image: Midjourney