$15B Later | Cool Pilots, Little Progress: Why Federal AI Still Doesn’t Work

Contrasting Phases: From AI Pilot Discussions in the Boardroom to Seamless Automation in Daily Operations.

The Illusion of Progress

Federal agencies must move beyond the illusion of progress created by pilots and PowerPoints and focus on automation efforts driving measurable outcomes, empowering better decisions, and aligning to mission goals like auditability, efficiency, and trust.

Too many AI and automation initiatives in government never move beyond the demo stage. They look impressive, check innovation boxes, and make it into briefings - but they rarely result in tangible improvements for federal employees or the citizens they serve. The root cause? A focus on tech for tech's sake, instead of outcome-oriented transformation.

From Spending to Scaling: Why the Gap Still Exists

Five-Year AI Spending: $15.05B Total, 86% for R&D, 14% for Operational Costs (Source: USASpending.gov).

Between FY2020 and FY2024, the U.S. federal government invested approximately $15.05 billion in artificial intelligence. Of that, $12.99 billion (about 86%) supported research and development (R&D), while only $2.03 billion (14%) was directed toward operational AI. Despite a growing recognition of AI’s importance, the numbers reveal a persistent bottleneck in moving from experimentation to execution.

This transition gap - often called the "valley of death" in innovation cycles - highlights structural challenges in scaling AI from research environments into live, mission-critical operations. Without dedicated implementation funding, agile procurement models, and strong governance, many promising initiatives risk stalling before they deliver real-world value.

A Better Way Forward: Outcome-Driven Automation

Success stories like the Defense Logistics Agency (DLA) show that it's possible to use AI and automation to fundamentally reshape how government works - when agencies start with outcomes, not tools.

Why Pilots Aren’t Progress

Pilots are important, but they often become the destination instead of the launchpad. Agencies frequently mistake a successful prototype or demo for implementation success. But real progress means:

• Deploying technology into live environments

• Enabling real users to benefit from it at scale

• Establishing supporting infrastructure and governance

• Driving organizational and cultural readiness

Don’t Automate the Wrong Things

“It’s not just about automating the way we do things, but completely redesigning our processes and thinking more about outcomes. That’s a really significant cultural shift in how we’ve historically thought about automation.” - Shawn Lennon, Deputy Director and Deputy CFO, DLA

Too often, automation efforts replicate broken processes instead of rethinking them. This accelerates inefficiency and locks in outdated workflows. Agencies must pause and ask:

• Is this the right process to begin with?

• Does this workflow still serve its purpose?

• Are we designing around the right decisions?

Transformational outcomes require redesigning workflows around the decisions that matter - not just optimizing routine tasks.

AI Is a Tool - Not the Strategy

Artificial intelligence is a powerful enabler, but it's not the end goal. Strategic success comes from improving how decisions are made: faster, with more confidence, and greater traceability.

To do this effectively:

• Embed AI into high-leverage decision points

• Enable human-machine teaming, not replacement

• Focus on accountability, not just automation

Case in Point: DLA’s Intelligent Approach

"You might have a 100-page contract, but I only need this field on page three and this field on page seven. AI can pull those data fields out, reconcile them to our accounting events, and present it to our accountant. Then they can push a button to hand it to the auditor. That’s how we’re going to pass the clean audit opinion." - Shawn Lennon, Deputy Director and Deputy CFO, DLA

At DLA, Deputy CFO Shawn Lennon shared how the agency is using AI and automation to support the Department of Defense's goal of a clean audit.

Rather than just digitizing legacy processes, DLA is:

• Rethinking workflows

• Using AI to extract and validate data

• Freeing up staff to focus on strategic work

• Enhancing audit-readiness and trust

Red Cedar’s Framework: Intelligent Transformation

At Red Cedar, Intelligent Transformation starts by identifying mission-aligned outcomes - what success looks like for our clients. From there, we work backward to design the technology, policies, and processes that will enable those outcomes. It’s not about plugging in a tool - it’s about building the right foundation to support scalable, trusted impact.

Our 5Ps framework ensures changes are:

• People-centered

• Policy-aligned

• Process-aware

• Partner-supported

• Platform-enabled

We apply Enterprise Architecture (EA) tools and techniques ensuring every AI or automation effort aligns with the mission and scales effectively. Drawing on DODAF and TOGAF, our team connects strategy to delivery by assessing current and future-state architecture, mapping decisions to systems, and designing modular, interoperable capabilities.

We also apply AI within our EA practice itself, leveraging automation to generate architecture diagrams, monitor compliance, and continuously optimize decisions and performance. From cloud optimization to real-time EA dashboards, we use AI to make enterprise architecture more dynamic, scalable, and adaptive.

When we help federal clients implement AI or automation, we start by asking:

• What decision is being made?

• What outcome are we trying to improve?

• What barriers stand in the way?

Only then do we architect a solution that fits - not just technically, but organizationally.

From Compliance to Capability

One of the most powerful ideas in the DLA example is that audit-readiness is more than a compliance box - it can be a strategic driver. When automation is aligned to goals like transparency, traceability, and accountability, it improves:

• Oversight and internal controls

• Operational resilience

• Trust from stakeholders

• Decision-making clarity

Final Thoughts | Build What Works, Not Just What Wows

AI and Automation that works isn’t flashy. It’s trusted. It’s used. It’s embedded into workflows and decision-making processes. And it scales because it delivers value where it matters most.

Federal agencies must stop mistaking activity for achievement. Pilots and PowerPoints have their place, but they’re not the finish line. Let’s shift the conversation from innovation theater to real impact.

That means:

• Starting with outcomes

• Questioning legacy processes

• Embedding AI in support of better, faster, and more accountable decisions

Only then will automation efforts move from optics to outcomes.

For more insights on transitioning from inefficiency to innovation in financial management, you can refer to the video featuring DLA Deputy CFO Shawn Lennon at Fedscoop.