Stop Building Software Factories. Start Building Founders.

How leadership churn is undermining defense software efforts

The Department of War doesn’t have a technology problem. It has a leadership continuity problem.

For more than a decade, the DoW has tried to replicate Silicon Valley’s innovation engine through software factories. Some acted as system integrators. Others focused on rapidly integrating commercial products. Many operated somewhere in between.

We copied the tooling. We copied the vocabulary. But we missed the one thing that makes Silicon Valley work: ownership.

The most innovative companies operate in “Founder Mode”—the people who start a product stay close to it long enough to see results and remain accountable for whether it makes a difference in the real world.

Founder Mode isn’t swagger, ego, or aimless micromanagement. It’s ownership in practice.

Owners measure outcomes in production: changes in behavior that drive mission impact. Not deployment frequency. Not green dashboards.

In agile development, knowledge lives in people, not in process. The DoW built software factories—and then rotated the knowledge out of them.

Rotation Breaks Products

DoW management culture is optimized for rotation and delegation. We reinforce it with familiar advice: hire good people and get out of their way. Manage through layers. That works when you’re executing a mature process. It breaks down when you’re building something new inside a resistant bureaucracy.

I’ve seen two failure modes.

The first is over-delegation: leaders too far removed from the work to help teams navigate acquisition constraints, security reviews, legacy integration, and cultural resistance.

The second, more common, is the opposite: leaders deeply involved but without product competence. Presence without understanding produces noise, not better decisions.

In the DoW, the obstacle is often the system itself. Teams need a leader who can navigate the institution on the product’s behalf.

Founder mode requires presence and competence.

The problem becomes clearer when we look at how digital products mature. Many digital products take five years to deliver impact. Agile doesn’t change that timeline. It shortens the feedback loop.

Teams can ship early, learn in production, and iterate—sometimes thousands of times—before the product provides value. But feedback only compounds if the same leaders are there to absorb it. Most DoW digital products have two or three leaders before they mature.

Each rotation resets context. Institutional memory walks out the door. Priorities shift. The organization stays busy but keeps starting over.

Weak measurement makes the problem worse. Some factories measure nothing meaningful. Others track delivery metrics—lead time, deployment frequency, mean time to restore. Those metrics matter but they are not mission impact.

The real question is whether they delivered outcomes. Did satellite operations planning time decrease? Did targeting accuracy improve? Did mission failure rates change?

If no one stays long enough to own those outcomes, they don’t get measured seriously. What isn’t owned defaults to activity. When leaders rotate every two years, accountability evaporates with them.

Agile without continuity is worse than the waterfall model we claim to be escaping. Waterfall assumes a stable plan. Agile assumes you will be wrong and need to learn. Learning requires memory and memory requires leaders who stay.

What Founder Mode Looks Like

If rotation is structural, then ‘Founder Mode’ must be an intentional choice to own the product and remain accountable for mission impact.

I co-founded Kessel Run, the DoW’s first software factory. But I didn’t start at the top. I started on a product team. For six months, we built our first application, shipped it to production, and iterated based on user feedback. Then I moved deliberately through portfolio and ecosystem responsibilities, learning in production alongside experts.

That progression mattered. It gave me the context to lead without guessing.Government does not grow product leaders by accident. If we expect leaders to stay longer, we also have to accelerate how they learn. Cohort-based learning, on-the-job apprenticeship, and deliberate exposure to product decisions are not nice-to-haves, they’re prerequisites.

Presence in the details isn’t micromanagement. It’s how leaders see whether their decisions work. When leaders understand the backlog and the mission context, trade-offs become sharper and drift becomes visible. When that leader stays long enough to see the results, accountability becomes concrete.

The product either has impact, or it doesn’t and the leader owns the outcome.

Build the System for Ownership

The DoW can’t rotate its way to durable digital capability. We assign five-year problems to leaders who rotate in two or three years. That mismatch guarantees resets.

Five-year products require five-year ownership.

That will require structural reform. The current assignment system produces generalists. Digital product ownership requires depth—specialized tracks, leaders outside the traditional command timeline, and a different way of valuing product expertise.

Leader development must change too. Move leaders from classrooms onto product teams. Let them live in the backlog and see what breaks in production. Build deliberate pathways from product to portfolio to ecosystem responsibility so leaders understand what they’re steering before they take the helm.

Finally, change what we reward.

Right now, the system rewards movement. Tours end on schedule. Contracts are awarded. Responsibility passes on.

If success means activity, leaders will optimize for activity. If success means measurable mission impact, leaders will optimize for outcomes.

Founder Mode is not about founders. It’s about ownership: leaders who stay close to what they start and remain accountable for what it becomes.

Stop trying to scale technology alone.

Start scaling leadership.