Thinking like a CTO

Because now we all need to

I ‘wrote’ this article on walk around Alexandra Palace today, and had Claude turn it into something coherent. With apologies to those who can smell the AI, but the thoughts (for good or ill) are my own.

---

A common refrain is that software engineers are becoming “orchestrators” of AI agents, that we’re managing teams of bots instead of writing code.

I think we need to go further.

We have to become leaders, not just managers.

Managing is about getting the thing done.

Leadership is about why we’re doing the thing in the first place.

That’s CTO thinking.

The Two Pillars of CTO Thinking

Vision. Not merely a probable future. A possible one. Something you can see clearly enough to align humans (and increasingly, AI agents) around it.

Heuristics. You can’t know exactly how you’ll get there. But you can steer by your values. For me, the most powerful heuristic has been optionality—the right to change your mind later.

Because every software decision is an economic trade-off.

Dollar today vs. ten dollars tomorrow.

Survival vs. investment.

Ship it fast vs. make it easy to change.

Even if you’re not trying to make money—even if you’re building for a charity or a cause—this trade-off remains. You’re balancing immediate needs against future value.

The Old Divide is Collapsing

These decisions used to live at the leadership level. Individual contributors could avoid them.

Or they’d fight a losing battle—”I know this will create tech debt, but my PM is pushing me to ship”—without understanding the full economic picture.

But AI is forcing this tension down to the IC level.

I saw this recently in a group I work with. An engineer new to Cursor burned through her token limit in days. She was using it to write tests, then arguing with the model about the results. Tokens gone.

So she had to make a CTO-level decision: Is this tool delivering enough ROI to justify upgrading?

She decided yes.

That’s CTO thinking. At the IC level. In real-time.

The Disconnect

This collapse is no bad thing.

One of the biggest frustrations of my career has been the disconnect between engineering and business.

Engineers not understanding, or shying away from, the economic pressures.

Leadership not understanding that the value isn’t just “features shipped”—it’s software that’s easy to change.

That gap created battles. Drama. Misalignment.

Now AI is forcing us to close that gap.

We’re all mini-CTOs now. We must ask: What’s the ROI of this approach? What’s the value of this tool? How do I balance speed against optionality?

Why This Makes Us Richer

Vision. Heuristics. Economic trade-offs.

That’s what thinking like a CTO means.

And it’s a skill set we all need to learn now. Not because AI demands it (though it does), but because it makes engineering whole again.

Economically, yes.

Functionally, yes.

But perhaps most important: spiritually.

I requires ask not just how, but why.

And there is no greater question we can ask.