Data-Driven Defense: The Army’s AI Era

Original Broadcast 7/20/2025

Presented by EY

Two leaders at the forefront of government transformation—George Kaczmarskyj, Principal for Government and Public Sector at EY and Leonel (Leo) Garciga, Chief Information Officer of the U.S. Army—share practical insights on the future of automation, data, and technology in the federal landscape. Together, they lay out a blueprint that combines commercial best practices with agency mission needs, all centered around one core principle: rethink the process before applying the technology.

Don’t Automate a Mess

George Kaczmarskyj kicks off the conversation by addressing a common pitfall: jumping straight to automation without evaluating the underlying process. “Don’t automate a mess,” he says—a phrase that headlines a recent piece he’s written. He’s seen it firsthand in the commercial sector, particularly in financial services where automation was adopted early and enthusiastically. But in many cases, the excitement led to automating tasks for automation’s sake.

Organizations would apply tools without rethinking the processes they were meant to improve. “We tried to screw [technology] in just to say we automated two hours of a person’s time,” Kaczmarskyj recalls. That approach ultimately sub-optimizes the opportunity.

He explains that real value comes from stepping back, asking why processes exist, and then reimagining or reengineering them. Only then, he says, should automation be applied—“as the last part of the mile.”

Another misstep Kaczmarskyj sees is over-reliance on a single automation tool or language. “To the Python developer, everything looked like a Python problem,” he says. Instead, agencies should recognize that different patterns in manual work require different technologies.

To address this, Kaczmarskyj recommends expanding the “technology aperture”—choosing from a broader landscape of automation tools based on the pattern of work, rather than defaulting to a familiar solution. He emphasizes the importance of process-first thinking and smart solutioning that considers the best fit for each specific task.

Kaczmarskyj also draws parallels to past government modernization efforts. Cybersecurity, for example, learned to “build in” protection rather than bolt it on. Digital transformation emphasized evaluating business processes before adding digital tools. “The solutions are similar,” he says, but automation can be even harder because the toolset is often fixed. The challenge is committing to that process of reimagination.

When asked about long-term planning, Kaczmarskyj identifies Year Five as the most difficult stage for agencies to envision—but also the most important. By that point, he says, organizations will be managing thousands of automations, hundreds of systems, and tens of thousands of stakeholders.

“If we don’t think about the consequences… we’ll create unnecessary technical debt,” he warns. To prepare, agencies must consider not only tools and implementation, but also the governance, functions, and job roles needed to support scaled automation.

Ultimately, Kaczmarskyj sees innovation as a core capability that must be built into the automation journey itself. It’s not just about tools—it’s about culture, mindset, and sustainability.

Inside the Army’s Data Revolution

Leo Garciga brings the military perspective to the table, detailing how the U.S. Army is tackling data transformation, cloud migration, and AI adoption. Speaking just after the Army’s second annual Data Summit, Garciga shares that attendance has jumped 980% over the previous year.

The Summit is the Army’s chance to take stock of where it is in its data story. According to Garciga, it’s the first time the Army has had “reps and sets” on major data initiatives and functional leaders actively using data across commands. New roles like Chief Data Officers are helping mature the environment and accelerate change.

A key driver of this growth is the Army’s effort to democratize access to data and tools. Platforms like Vantage are enabling soldiers and civilians to answer their own questions and build their own solutions. “It’s mostly organic at this point,” Garciga says, with government personnel leading the way rather than contractors.

Garciga references what the Army informally calls the “automation or die” memo—a directive from the Secretary of the Army to identify and automate high-volume manual processes. The rise of low-code/no-code platforms is making this possible, helping users build solutions quickly and at scale.

AI is also advancing rapidly. In just 72 hours, over 5,000 people joined the Army’s new enterprise AI workspace. That number has since grown to 19,000. The platform offers access to commercial large language models under strong guardrails. Users range from novices doing advanced search to experienced developers writing code and launching pilots.

Another key transformation is the Army’s move to the Joint Warfighting Cloud Capability (JWCC) contract. The Army is now its largest user, accounting for 62% of all workloads. Garciga explains the shift brings better buying power, streamlined operations, and clarity across the force.

He also details progress toward continuous authority to operate (C-ATO). After 24 months of groundwork, three Army programs now operate under continuous authorization—a milestone Garciga sees as critical to scaling innovation across programs and commands.

Ultimately, Garciga believes the Army’s success depends on cultural change. By making tools widely available in a safe environment, leaders are encouraging experimentation and learning. “Not everyone’s an expert,” he says, “but like learning to drive a car, you’ve got to get behind the wheel.”

That culture of permission—combined with commercial-first thinking, simplified platforms, and strong training—is enabling the Army to modernize with speed and purpose.