The Last Mile of Pentagon Reform Runs Through the Brigade

A kill chain measured in seconds cannot depend on processes measured in weeks or months.

以秒为单位的杀伤链不能依赖以周或月为单位的进程。

The Pentagon recognizes that drone proliferation and AI-enabled lethal targeting are reducing decision cycles and increasing the premium on speed.

五角大楼认识到无人机的扩散和人工智能支持的致命目标正在缩短决策周期并增加对速度的重视。

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.

2025 年无人机和采办改革以及 2026 年人工智能战略呼吁加快采购速度、减少集成障碍、开放架构、持续实验以及更多地获取新兴技术。

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.

然而，我和第 82 空降师第 1 旅战斗队 (1/82) 的战友在部署支持史诗狂怒行动之前和期间的经验表明，这一改革议程在战术方面存在缺陷。

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.

尽管五角大楼和作战司令部在加速杀伤链方面取得了所有进展，但那些处于尖端的部队在整合无人机、瞄准软件、网络和火力系统方面仍然面临着延误。

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.

旨在加速创新的政策并没有始终如一地给予旅级部队在作战期间组装和调整综合杀伤链所需的权力、技术支持和开发权限。

The Army has the technology to compete, but institutional latency constrains tactical formations’ ability to adapt.

陆军拥有竞争技术，但制度滞后限制了战术编队的适应能力。

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 .

战术指挥官需要权力、访问权限和技术支持来控制和调整自己的杀伤链，同时保持目标权威和人类对武力使用的判断。

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.

缩小这一差距需要将采购权下放给旅指挥官，将整合权下放给师指挥官，为旅提供专门的技术支持，并让战术编队直接参与人工智能开发。

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.

我们学到了什么 构建数字杀伤链 杀伤链是检测和确认敌方目标、将信息传输给射手、与目标交战并评估效果的过程。

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.

全数字杀伤链将在共享计算环境中将传感器和射手联网，使人工智能能够快速识别目标并自动将其与适当的射击系统配对，同时让训练有素的操作员负责致命决策。

This ideal is still far from the reality experienced by soldiers today.

这种理想与当今士兵所经历的现实仍然相去甚远。

For example, 1/82’s legacy kill chain process involved a sensor, such as a drone, using full-motion video to observe possible targets.

例如，1/82 的传统杀伤链过程涉及无人机等传感器，使用全动态视频来观察可能的目标。

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.

初级伞兵接受的训练有限，在无人机杀伤链方面的经验还不到一年，他们观看了视频，在笔记本上写下了网格，并通过多个物理指挥所将它们作为单独的火力任务口头传输。

Every manual handoff introduced delays into the process, forcing capable and motivated paratroopers to spend precious time acting as data couriers and analysts.

每次手动交接都会导致流程延迟，迫使有能力和积极性的伞兵花费宝贵的时间充当数据信使和分析师。

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.

在我们的集成和测试工作中引入的快速射击通道等可能的解决方法通过绕过中间指挥所将任务直接发送给射击单位来缩短杀伤链。

However, the underlying process remained unchanged as these nets still relied on voice communications, manual data transfers, and analog systems.

然而，底层流程保持不变，因为这些网络仍然依赖于语音通信、手动数据传输和模拟系统。

Additionally, removing intermediate steps concentrated more traffic at the receiving end, increasing the burden on personnel processing fire missions and network congestion.

此外，消除中间步骤会在接收端集中更多流量，增加处理消防任务的人员负担和网络拥塞。

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.

我们发现传感器、云服务、瞄准软件和火力引导系统是为了不同的目的、在不同的时间线上、使用不同的技术接口而开发和部署的。

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.

连接它们需要定制的网络配置和数据链路、不断的故障排除和重复的测试，然后数据才能可靠、干净地从一个系统流到另一个系统。

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.

但是，尽管投入了大约 3,500 个伞兵工时进行集成、测试和实验，而不是发明，1/82 仅实现了部分数字化的杀伤链。

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.

无人机视频和目标数据可以路由到 IL5 云环境，但由于集成和转换问题，操作员仍然必须检查和纠正数据。

They also had to manually transmit targeting packets to lethal fire direction systems because systems could not reliably exchange information.

他们还必须手动将目标数据包传输到致命的火力指挥系统，因为系统无法可靠地交换信息。

These technical workarounds left the process too slow for combat operations in Central Command.

这些技术解决办法使得中央司令部的作战行动过程变得过于缓慢。

1/82 also had access to Palantir’s Maven and successfully integrated some, but not most, of its sensors and digital systems.

1/82 还可以访问 Palantir 的 Maven，并成功集成了部分（但不是大部分）传感器和数字系统。

However, even when we established links, data cleanliness and translation problems persisted.

然而，即使我们建立了链接，数据清洁度和翻译问题仍然存在。

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.

对于此集成工作，缺乏直接的现场工程支持，导致排除这些链路故障或建立其他链路的速度缓慢且耗费大量人力。

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.

需要明确的是，1/82只是一种以其独特的系统、人员和作战需求组合运作的编队，因此它的经验并不能证明每个战术编队都会遇到相同的障碍。

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.

例如，第 4 和第 25 步兵师在大力支持的原型设计和实验工作中与数十个行业合作伙伴和多个陆军项目办公室合作，第 25 步兵师展示了完全数字化的火力链。

But those efforts benefited from purposeful funding, direct and robust technical support, and access to development partners that tactical formations do not routinely possess.

但这些努力受益于有针对性的资金、直接和强有力的技术支持，以及战术编队通常不具备的发展合作伙伴的机会。

They also occurred largely in training and experimentation environments, where integration can be bounded in ways that operational adaptation cannot.

它们也主要发生在培训和实验环境中，在这些环境中，集成可能会受到操作适应所无法限制的限制。

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.

快速而成功的技术适应部分取决于更快地签订技术服务和必要的软件和设备合同、更容易的系统集成、开放式架构和持续的实验。

For 1/82, connecting components that worked independently required continuous data engineering, network reconfigurations, and testing.

对于 1/82，连接独立工作的组件需要持续的数据工程、网络重新配置和测试。

Contracted engineers could help with that work, but hiring them would not, by itself, resolve delays in obtaining integration approvals.

合同工程师可以帮助完成这项工作，但雇用他们本身并不能解决获得集成批准的延迟问题。

More broadly, the challenge faced by 1/82 was translating that strategic intent into authority and support at the brigade level.

更广泛地说，1/82 面临的挑战是将战略意图转化为旅级的权威和支持。

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.

控制资金的陆军司令部限制其旅优先级的释放，根据所涉及的系统由不同的当​​局进行整合批准，并且技术支持只能间歇性地提供。

The consequence was a gap between enterprise-level reforms and the capabilities available to tactical formations in combat.

其结果是企业层面的改革与战斗中战术编队的可用能力之间存在差距。

Many of the enterprise-level controls serve legitimate purposes.

许多企业级控制措施都有合法的用途。

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 .

采购监督保护公共资金并限制对个别供应商的依赖；网络安全和系统审批流程保护我们的网络；当数据走向致命影响时，针对性的保障措施可以降低出错的风险。

The challenge is balancing those assurances against the operational risk created when adaptation and approval timelines lag behind battlefield requirements.

面临的挑战是在这些保证与适应和批准时间表落后于战场要求时产生的运营风险之间取得平衡。

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

以下四项建议将帮助战术编队快速适应和改进其杀伤链，同时保留必要的保障措施。

The July 2025 drone reforms delegated greater authority over small drones to lower echelons and emphasized faster modification by frontline personnel.

2025 年 7 月的无人机改革将小型无人机的更大权力下放到了较低梯队，并强调一线人员更快地进行改造。

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.

我们的经验表明，同样的方法应该扩展到将这些系统连接到有效杀伤链所需的商业软件和服务。

For 1/82, the immediate constraint was delegated authority to use available funds.

对于 1/82，直接的限制是授权使用可用资金。

Curtailed financial authorities limited the brigade’s ability to contract for services and purchase commercially available equipment and software.

财政当局的削减限制了该旅签订服务合同以及购买商用设备和软件的能力。

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.

即使是少量采购人工智能软件系统和网络工具，也需要以财政季度为单位衡量采购时间表，从而延迟了现有技术的整合，并阻止战术指挥官根据战斗节奏调整杀伤链。

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.

为了使财务权力与战场现实保持一致，旅指挥官需要为其整合优先事项提供资金，并有权批准如何使用这些资金。

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.

这应包括个人购买高达 50,000 美元的商用设备和软件以及高达 100,000 美元的个人技术服务合同，这是非常合理的初始限额。

That delegation would remain bound by existing fiscal controls , competition requirements , cybersecurity standards , and applicable safety and fielding requirements .

该授权仍将受到现有财政控制、竞争要求、网络安全标准以及适用的安全和部署要求的约束。

Where practical, purchases should also include open interfaces and government data rights to reduce dependence on a single vendor and preserve future integration options.

在可行的情况下，采购还应包括开放接口和政府数据权利，以减少对单一供应商的依赖并保留未来的集成选项。

This authority would help close time-sensitive tactical integration gaps, while enterprise acquisition remains responsible for enduring programs and common architectures.

这一权力将有助于缩小时间敏感的战术集成差距，而企业采购仍然负责持久的项目和通用架构。

Larger software requirements and sustained engineering support would require separate funding and approval arrangements.

更大的软件需求和持续的工程支持将需要单独的资金和审批安排。

The January 2026 Artificial Intelligence Strategy and other initiatives referred to in this article call for fewer integration barriers.

2026 年 1 月人工智能战略和本文提到的其他举措呼吁减少集成障碍。

For 1/82, connecting tactical data environments, fires systems, and external data services still required multiple high-level approvals.

对于1/82来说，连接战术数据环境、火力系统和外部数据服务仍然需要多个高层批准。

Even minor data-routing adjustments required weeks of coordination.

即使是微小的数据路由调整也需要数周的协调时间。

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.

在某些情况下，1/82 必须与部门级办公室协调解决战术整合问题，从而增加了活跃战区中出现的问题的距离和延迟。

That delay creates operational risk, but the legacy approval process also addresses legitimate concerns.

这种延迟会带来运营风险，但遗留的审批流程也解决了合理的问题。

Connecting previously separate systems can introduce cybersecurity vulnerabilities or degrade data integrity.

连接以前独立的系统可能会引入网络安全漏洞或降低数据完整性。

Those risks become especially consequential when digital workflows feed targeting and weapons employment.

当数字工作流程支持瞄准和武器使用时，这些风险变得尤其严重。

Any faster approval process should therefore preserve technical review and established targeting safeguards.

因此，任何更快的审批流程都应保留技术审查和既定的目标保障措施。

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.

因此，作战链中的第一个师级指挥官应在既定的网络安全和风险管理标准范围内获得授权，以批准战术系统连接和有条件的操作授权修改。

Previous efforts to improve the authorization process largely focused on accelerating or streamlining it at the enterprise level.

以前改进授权流程的努力主要集中在企业层面上加速或简化授权流程。

This proposal takes a different approach by moving bounded authority for time-sensitive tactical integrations closer to the operational problem.

该提案采用了不同的方法，将时间敏感的战术整合的有限权力移至更接近作战问题的位置。

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.

将军总部可以汇集作战、情报、消防、网络安全、法律、通信和技术专业知识，以评估拟议连接的风险以及延迟连接的操作后果。

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.

根据这一授权，批准将需要实施后技术审查，并有范围和时间限制，并具有明确的接口、监控要求、回滚程序和强制性后续审查。

This approach allows commanders to accept operational risk without turning temporary adaptations into permanent exceptions to cybersecurity standards.

这种方法允许指挥官接受作战风险，而无需将临时调整转​​变为网络安全标准的永久例外。

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.

授权可以在几天而不是几周或几个月内评估和批准有限的修改，同时保留监督并使集成权限与运营责任保持一致。

Tactical integration requires both delegated authority and dedicated technical expertise.

战术整合需要授权和专门的技术专长。

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.

五角大楼的 2025 年 11 月采办转型战略区分了技术就绪性和集成就绪性，并承认缺乏资源充足的途径来将商业技术快速集成到集成杀伤链中。

Our brigade experienced that gap directly.

我们旅直接经历了这种差距。

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.

伞兵在部署前投入的大约 3,500 个小时中，很大一部分用于由未履行正常作战职责的士兵进行的系统集成和数据路径故障排除。

In some cases, integrations had to be repeated across hundreds of platforms rather than applied across an entire fleet, an incredibly labor-intensive process.

在某些情况下，必须在数百个平台上重复集成，而不是在整个机群中应用集成，这是一个极其耗费人力的过程。

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.

专业工程支持时断时续，而且往往难以获得，迫使我们依赖非正式关系，有时伞兵需要等待数天或数周才能恢复测试。

Faster purchasing and approval address only part of the integration problem.

更快的采购和审批只能解决部分集成问题。

Creating interfaces, translating data, configuring networks, and troubleshooting failed connections still require engineers who can work alongside operators.

创建接口、转换数据、配置网络以及对失败的连接进行故障排除仍然需要能够与操作员一起工作的工程师。

Dedicated direct-support technical teams should therefore support brigades before and during combat operations.

因此，专门的直接支援技术小组应该在战斗行动之前和期间为旅提供支持。

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.

他们不需要成为旅的建制，但他们应该投入人员和时间来直接支持旅指挥官的整合优先事项。

In the near term, these teams would rely on contracted engineers.

在短期内，这些团队将依赖合同工程师。

Over time, the Army should build a military and government civilian technical workforce with requisite system access and rights.

随着时间的推移，陆军应该建立一支拥有必要的系统访问和权利的军事和政府文职技术人员队伍。

They would engineer data links, automate targeting workflows, troubleshoot integrations, and continuously adapt digital kill chains as systems and operational requirements change.

他们将设计数据链接、自动化目标工作流程、排除集成故障，并随着系统和操作要求的变化不断调整数字杀伤链。

Embedded engineers would also strengthen assurance and reduce risk by identifying unsecure configurations and risky integrations before commanders accept them operationally.

嵌入式工程师还将通过在指挥官接受操作之前识别不安全的配置和有风险的集成来加强保障并降低风险。

Persistent technical support would make continuous integration and adaptation a routine brigade function.

持续的技术支持将使持续集成和适应成为旅的日常职能。

Tactical formations should be central participants in developing emerging AI capabilities.

战术编队应该是开发新兴人工智能能力的核心参与者。

Brigades can identify operational needs, contribute sensor data, and test capabilities with developers responsible for the underlying software and models.

旅可以确定作战需求，提供传感器数据，并与负责底层软件和模型的开发人员一起测试能力。

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.

目前，他们对这些开发人员的访问有限，限制了使用操作数据来完善目标识别模型或测试人工智能辅助火力方向工作流程等的机会。

They also struggle to access dependable warfighting AI tools on “sensitive but unclassified” networks, along with the experimentation programs and events that shape them.

他们还努力在“敏感但非机密”的网络上获取可靠的作战人工智能工具，以及塑造它们的实验计划和事件。

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.

这种脱节减缓了战场经验和人工智能开发之间的反馈循环，并使战术编队依赖于可能无法反映作战相关目标或环境的软件。

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.

Swarm Forge 和 Agent Network 等项目建立了一种将操作实验与快速人工智能开发相结合的方法，但它尚未在离散实验工作之外一致地形成战术编队。

Brigades should therefore have a formal role as continuous development partners, with regular and direct access to supporting developers.

因此，团队应该扮演持续开发合作伙伴的正式角色，并定期直接接触支持开发人员。

Participation should focus on testing and refining warfighting AI programs on approved sensitive but unclassified networks and providing operational data and feedback.

参与应侧重于在经批准的敏感但非机密网络上测试和完善作战人工智能程序，并提供操作数据和反馈。

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.

如果现有设备无法运行软件进程，则应为部队提供与云计算资源或更强大的计算设备的高带宽连接。

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.

访问将允许中央司令部生成的作战数据为模型细化和工作流程测试提供信息，帮助人工智能能力按照与战场适应相关的时间表发展。

The Pentagon’s recent drone, acquisition, and AI reforms are movement in the right direction.

五角大楼最近的无人机、采办和人工智能改革正在朝着正确的方向发展。

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.

然而，1/82 的经验表明，成功最终取决于这些改革是否改变了战术边缘的条件，在战术边缘，编队应不断整合快速变化的技术，而这些技术尚未完全实现互操作。

Success also requires moving risk decisions to echelons that can weigh warranted safeguards against the operational consequences of delay.

成功还需要将风险决策转移到各个梯队，以便权衡必要的保障措施和延误造成的运营后果。

That requires bounded purchasing authority, integration decisions close enough to operations to matter, dedicated technical expertise, and routine participation in AI development.

这需要有限的购买权、足够接近运营的集成决策、专门的技术专业知识以及对人工智能开发的例行参与。

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.

最终，衡量五角大楼改革的标准应该是一个旅是否能够以战场的速度调整其杀伤链，而我们在这一指标上还达不到要求。

Lucas Ziller is a U.S. Army major and infantry officer with experience in conventional and special operations assignments.

卢卡斯·齐勒 (Lucas Ziller) 是一名美国陆军少校和步兵军官，拥有常规和特种作战任务经验。

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.

最近，他在美国中央司令部第 82 空降师第 1 旅战斗队服役，帮助将人工智能系统和自主技术整合到作战中。

He is currently a Ph.D. student in Public Administration at North Carolina State University.

他目前是博士学位。北卡罗来纳州立大学公共管理专业的学生。