What Most AI Programs Miss and Why National Security Cannot

Modernization inside the Department of Defense (Department of War) rarely fails due to lack of ambition. It stalls when organizations cannot move from intent to execution fast enough to meet mission demands. On the USSOCOM Enterprise Development, Application and Training (EDAT) program, Kentro’s team addresses that challenge by aligning technology sourcing, disciplined delivery and outcome-driven use of AI to accelerate mission impact through rapid prototyping. 

Across industry, artificial intelligence initiatives continue to have challenges. Research shows that roughly 80 percent of AI projects fail to deliver meaningful value. Pilots stall after initial experimentation, while others collapse under security, integration or governance challenges. In most cases, failure stems from applying technology without a clear path to outcomes. 

Our EDAT Team took a different approach

As Kentro launched its Estelle iLab earlier this year, demand grew quickly for AI-enabled demonstrations that support customer engagements. The ability to move from concept to working example became essential. Replit emerged as a practical tool for rapidly creating functional demonstrations helping customers visualize solutions early.  

Working with Technical Services and leadership, Kentro procured Replit licenses for the EDAT Team. That decision marked a shift from experimentation to applied execution. 

Within the EDAT environment, customers understand the outcomes they need but struggle to express those needs through written requirements alone. Traditional discovery cycles delays progress and increases rework. Rapid prototyping offers a way to replace abstract discussions with something tangible. 

Using Replit, the EDAT Team converted early requirements into functional prototypes within 48 hours. Customers clicked through workflows, explored features, and provided immediate feedback. Rapid prototyping clarified intent, resolved ambiguity early, and allowed teams to move forward with confidence. In practice, this approach shortened the discovery phase by an estimated four months, driving both cost savings and cost avoidance for the customer.  This process reduces the risk of a failed deployment and focuses the scope on real objectives and meaningful outcomes.  

AI supported the process without dictating it. The team uses Replit to accelerate ideation and structural design, then rebuilt solutions within secure government environments using approved application stacks and architectures. Developers applied security frameworks, validated code quality and ensured compliance before deployment. This approach preserved speed while maintaining trust. 

On one SOCOM initiative, AI-enabled rapid prototyping shortened development timelines by six weeks and saved the government $220,000. It enabled the delivery of a minimum viable product under a timeline the customer had previously considered unrealistic. 

This outcome highlights why many AI initiatives fail and why our EDAT Team’s approach succeeded. Broad transformation efforts often attempt to do too much at once. We focused on a specific workload where AI could deliver immediate value. The team anchored every decision to mission outcomes and applied technology only where it fit. 

They brought us an impossible problem set,” said Richard Marshall, Kentro’s Director, SOCOM. “They asked us to get to a minimum viable product in under six months on something they had been trying to complete for more than four years. We delivered on that promise, and it changed how they view what’s possible.

Rapid prototyping is reshaping how customers engage with the development process and how teams align around shared outcomes. Instead of debating requirements in isolation, stakeholders collaborate around working systems. That collaboration strengthened trust and accelerated progress. 

The lesson from EDAT is clear. AI delivers value when teams apply it with discipline, intent, and focus on outcomes. Rapid prototyping provides tremendous value as it shortens understanding and accelerates alignment. 

On EDAT, technology sourcing led to execution, execution led to trust and trust led to results. That alignment turned rapid prototyping into a catalyst for mission progress and demonstrated how AI, used deliberately, can deliver where most initiatives fall short.