What happens inside the Pentagon is special. How it gets done is not.
From a business systems perspective, the Pentagon operates in essentially the same manner that any other multinational conglomerate does.
Yet, the department has long seen its requirements, even those for everyday business operations, as unique. To meet those supposedly one-of-a-kind needs, it has been willing to foot a hefty upcharge to build highly customized software applications. Overengineered and underpowered since their inception, many of these services are now unwieldy productivity blockers that lag years behind offerings available to the private sector that are simultaneously powerful, usable, and cost-effective.
With few unclassified exceptions, the national security community needs the same corporate software as the rest of us. But that’s not something it’s readily admitted to until recently.
Commercial-first is no longer just a suggestion
Software is the perfect place to begin practicing the “buy before build” mantra: buy what you can, and build only what you must. If acquirers believe there is no suitable commercial offering, they are required to produce documented market research to prove that no existing product can meet the requirement, even with reasonable modifications. Law and policy are unambiguous at this point: commercial must be the default, not the afterthought.
Silicon Valley and Washington have never been more closely aligned. The federal government’s senior-most appointed roles are filled with executives from the private sector, hailing from deep tech unicorns, venture capital, and investment banking. These leaders bring exquisite knowledge of what it really means to move fast, and how to replicate it. They’ve been wielding it to great effect. Commercial awards across the enterprise have quietly demonstrated something long suspected: when the government actually buys what industry already built, outcomes improve and timelines compress.
Bespoke builds create bespoke problems. It’s the primary reason the Pentagon has a deep dependence on Frankensteined platforms that require FTEs to operate and maintain, not to mention years of expensive contract modifications that sometimes don’t even yield a minimum viable product. There is no shortage of cautionary historical examples: DIMHRS (a ~$1 billion human resources platform that was never delivered), DTS (a ~$0.5 billion travel booking site with almost $10 billion per year in fees), and AHLTA (~$4 billion medical records software, not to mention its follow-on, which is expected to reach ten times that figure). Ballooning costs aside, users will freely admit that these platforms are borderline unusable and frequently offline.
A $250M task management platform is today’s $9,609 wrench
While bespoke software may have a specific place in defense, it’s certainly not in unclassified business systems. In early 2026, the Pentagon formally acknowledged that its two still-quite-nascent (circa 2019) task management platforms, ETMS2 and CATMS, are falling short, and recently issued a call for solutions for a single modernized system capable of being operated by more than 150,000 department-wide users. The price tag of those predecessors is eyewatering. The cumulative cost through 62 modifications over five years for ETMS2 alone is more than $250 million.
CATMS and ETMS2 are how defense staffers receive, review, respond, and complete a great deal of their work today. These two platforms serve as the execution infrastructure and the backbone of Pentagon process. Anyone exposed to these tools recognizes the importance of their existence but obvious shortcomings in their usability. They require users to sift through volumes of irrelevant subject matter to determine organizational equity, manually retype routing templates, play games of telephone with time-sensitive information, and slap together disparate comment resolution matrices at multiple echelons of the review process. Critical tasks are often delegated to users after their due date because every handoff injects latency, they are regularly reassigned due to incorrect routing, and they are frequently missed due to the sheer volume of administrivia involved. Handjamming data from one place to another when the most basic software can do the same instantaneously is effectively treating the defense workforce like human APIs.
This enterprise replacement solicitation is a bold attempt to bring the department into the 21st century. The intent with the new platform, and many other forthcoming solicitations, is noble: invest in tools that make people more capable. Since the department granted its users access to world-class generative AI tools — several years after the rest of the world — productivity and satisfaction have undoubtedly increased.
Arming the people behind the arsenal
Revamping the way the enterprise routes and responds to taskers appears to be a straightforward modernization effort. But the Pentagon has failed tests like this one before.
The call for solutions rightly prioritizes software that is “mature, scalable, and secure” without seeking custom builds and government-unique systems. It asks for software that can be fielded immediately, requires scale and speed where previously just a prototype would’ve sufficed, and specifies a commercial off-the-shelf (COTS) model in order to prevent data siloes, ensure interoperability, and eliminate the risk of vendor lock.
Accordingly, a successful award would be expected to heavily weight aspects like platform maturity, velocity of fielding powerful new features, degree of AI native capabilities, obsession with user experience, and deployment to multiple classified networks, built by companies that put their own skin in the game.
Generative AI is a major differentiator and an overdue efficiency gain in the new software suite available to the defense workforce. But not all AI capabilities are created equal — some platforms were engineered with AI as the bedrock while others bolted it on later as an afterthought.
The vendors best positioned to serve the Pentagon’s needs have engineered their platforms with AI as a foundational capability, not those retrofitting it onto legacy architecture. Market forces incentivized a generation of modern software companies to build with AI tools and design their platforms for native AI capability from the start. These companies iterate faster, respond to user needs in real time, and deliver software that prioritizes usability as well as security. Companies shaped by the AI arms race have a culture of velocity embedded in their DNA. Legacy contractors bolting AI onto monolithic architectures never designed for it will never match the pace of AI-native successors built from the data layer up.
Software is the force multiplier
Business systems are the kit of service staffs, their command and agency counterparts, and the Office of the Secretary of Defense. The best software in the world is commercially available today, built and hosted domestically by companies that understand how modern organizations operate, and is already deployed to secure government networks.
The tools exist, policy mandates their adoption, and the workforce has been awaiting their arrival. The question is whether the department will procure another billion-dollar bespoke build that falls short and lags the state of the art over for its entire lifecycle — or if it will give the workforce and taxpayer what they deserve.
