Ch8 03: The CEO Who Wouldn’t Leave the Factory Floor#

Brandon Krieg had a problem. As CEO of Stash — a fintech company with millions of users — he was watching the AI revolution unfold and knew his product needed to evolve. Fast. The question wasn’t whether to weave AI into Stash’s financial coaching features. The question was how to make it happen before the window slammed shut.

He had the standard playbook. Hire an AI team. Bring in consultants. Spin up a task force, commission a strategy review, roll out a phased implementation plan. Any of those moves would’ve been reasonable. All of them would’ve eaten months.

Brandon went a different direction. He pulled together a tiny team — fewer than eight people — and started building the AI coaching feature himself. Not overseeing. Not reviewing. Building. Writing specs, testing prototypes, iterating on outputs. The CEO of a company with hundreds of employees, parked at a desk, doing the work of a product manager and an engineer.

From the outside, it looked like a demotion. From the inside, it was the most effective leadership decision he made that year.

There’s a phenomenon in organizations I call signal decay. When a CEO announces a strategic priority — “AI is our number one priority this year” — the message starts at full blast. Direct reports hear it loud and clear. They relay it to their teams, where it loses some volume. By the time it trickles down to the people who actually build things, it’s been diluted by competing priorities, refracted through departmental lenses, and softened by middle managers hedging their bets.

In a five-layer organization, a strategic signal sheds roughly thirty to fifty percent of its force at each layer. By the time it reaches the fifth layer, “this is the most important thing we’re doing” has faded to “this is one of several things we should probably get to when we have bandwidth.”

Brandon’s move — building the feature himself — bypassed signal decay entirely. He didn’t send a message about priorities. He was the message. When engineers saw the CEO sitting next to them, working the same codebase, attending the same standups, testing the same prototypes, nobody had to wonder whether AI was a real priority. The signal was unambiguous, transmitted by the most powerful person in the company in the most unmistakable format available: action.

There’s a second payoff that’s less obvious but equally big. When a leader builds the prototype, the organization can’t kill it through the usual bureaucratic antibodies.

Big organizations have immune systems. Any new initiative — especially one that challenges existing processes or threatens existing fiefdoms — triggers an immune response. Committees spin up to “evaluate.” Risk assessments get commissioned. Pilot programs get proposed and then studied to death. The immune system doesn’t reject new ideas with an outright “no.” It kills them through delay, qualification, and slow absorption into the existing machinery.

A working prototype built by the CEO short-circuits this immune response. You can’t “evaluate” something that already works. You can’t propose a “pilot” for something the CEO has already demoed. The prototype is a fait accompli — a concrete artifact that flips the conversation from “should we do this?” to “how do we scale this?”

I call this the results-first approach: build it, then convince. It’s ten times more effective than the traditional route — convince first, then build — because it swaps speculation for evidence. Every objection that starts with “but what if…” is answered by “here, look at it.”

Brandon’s team shipped the AI coaching feature in weeks, not months. It worked. Users responded well. And the rest of the organization — which might have spent months debating the approach — pivoted to support and scale what had already been proven.

But the ripple went beyond the specific feature. Something shifted in Stash’s culture. People saw the CEO willing to do the work, not just direct it. That signal — “I care about this enough to do it myself” — rewired how the entire organization approached execution. Meetings got tighter. Decision loops got shorter. The gap between “we should” and “we did” compressed.

This is the deepest value of leadership by doing. It’s not just about the specific deliverable — the prototype, the feature, the product. It’s about the cultural message baked into the act. When the leader does the work, every person in the organization receives the same signal at the same instant: this is what matters. This is how fast we move. This is the bar.

No memo, no all-hands, no strategy deck can transmit that signal with the same clarity.

I want to be precise about what I’m not saying. I’m not saying CEOs should spend all their time building products. There are genuine leadership duties — strategy, culture, talent, external relationships — that only the CEO can handle. Ditching those to write code would be a mistake.

What I’m saying is that there are moments — specific, critical junctures when a new direction needs to be planted, when the org’s immune system is fighting change, when speed matters more than process — where the most effective thing a leader can do is step off the stage and pick up a wrench.

These moments are rare. But they’re decisive. And the leaders who spot them — who have the judgment to know when their personal involvement will catalyze change, and the humility to do work that’s “beneath their title” — are the ones who build organizations that can sprint at startup speed even as they scale.

Guidance#

Ask yourself: is there a project in your organization right now that everyone agrees is important but that isn’t moving fast enough?

If yes, ask why. Resource problem? Talent gap? Or is it a signal problem — the organization isn’t moving because it isn’t sure the priority is real?

If it’s a signal problem, here’s the most powerful lever you own:

1. Pick one project. Not three. Not five. One. The single most important strategic shift for the next quarter.

2. Join the team. Not as a reviewer or sponsor. As a working member. Show up to standups. Review the work. Contribute directly. Your presence is the signal.

3. Set a deadline in weeks. Not months. Weeks. The compressed timeline forces the team past the usual bureaucratic drag and into building mode.

4. Ship something. It doesn’t need to be polished. It needs to be real — a working prototype, a functional demo, a tangible artifact that proves the concept. Polish comes later. Proof comes now.

5. Step back. Once the prototype exists and the direction is locked, return to your normal leadership role. Hand the project off with a clear mandate and the resources to scale. Your job was to start the engine. Their job is to drive.

The most powerful form of leadership isn’t giving orders. It’s setting an example. And the most powerful example is the one where the leader does the work.