Skip to content

Finding new challenges after decades of engineering

Much as I talk about how the current generation of AI tooling is missing the point, and it is, I still think the technology is incredible.

I’m doing projects now, casually, that I could not have done by myself even as recently as two years ago. That’s especially true of the creative and productivity apps I’ve been designing lately.

The spots where I still have to step in are exactly the places you would expect: precision work, careful testing, taste, judgment, and the pieces of an app where the delta between “close enough” and “correct” needs attention.

And honestly, that’s fine. That’s the work I still enjoy doing by hand.

I’ve never had this much diversity in the kinds of projects I can take on as a consultant. Nothing scares me anymore. My productivity is high, and I feel creatively satisfied in a way that takes me back to the beginning of my career. The excitement of it, the boundless greenfield of it all. The opportunity to rethink the tech and the culture... again.

It reminds me of my original guiding philosophy.

That I am an agent of change.

Because change is the thing I thrive on as a knowledge worker.

What I was tired of

In late 2023, just before the first public coding agents hit, I was at a point in my career where I was genuinely questioning whether I wanted to be a software engineer anymore.

The more I think about it, the more I realize that creating code was never really the problem. It wasn’t the people, usually. And, even though I have my criticisms of the way enterprise teams are managed these days, it wasn't that either.

What I was tired of was the mechanical process of physically writing code by hand.

When I first started, that was interesting. It was challenging. It felt like a puzzle. But over time, I fell into a rut where the work stopped feeling creative or difficult enough to wake up the good parts of my brain.

Turn the failure states into art

So when I started playing with AI, first graph networks, then computer vision and diffusion, then agents, what I wanted to know was simple:

How do you turn the failure states into art?

Obviously, that led to me pushing myself and my machines harder than I ever had.

And I never really stopped.

This is the good stuff.

I can do this all day.