On the Existential Crisis of Software Engineers

The software development world is changed. I feel it in the water. I feel it in the earth. I smell it in the air.
/ Josh Larson / techaiweb-developmentsoftware

Like many other software engineers tuned into the AI landscape, I’ve been feeling really weird about our career paths for the past few months.

I couldn’t quite put my finger on the source of the feeling, but then Rachel posted:

Bingo.

As a senior-level software engineer in February 2026, I rarely write code by hand anymore. It feels weird to admit that!

Why it feels weird

I was talking to my buddy Corey this week about how someday we’ll be looking back on the things we built in the pre-AI world. We’ll probably laugh and say, “Back in my day, I wrote that code by hand!”

But it’s true:

  • I built an entire SaaS side-business by hand. GitHub CoPilot was starting to become a thing by the time I sold it, but nothing near the level of agentic coding that exists today.

  • Flareact was a fun experiment I built in a week with my own fingers. It’s what got me my job today, and I owe most of my career wins of the past half-decade to that project.

  • Hydrogen and Superflare are examples of meaningful work and passion projects that have come together completely by manual coding.

So what’s left of my identity as a software engineer?

My previously-laid plans

Superflare seemed like the wisest place to invest my (extremely-limited) spare time and energy. A modern, full-stack framework built on Cloudflare Workers. I was going all-in on excellent developer experience, ergonomics, and fluent code that rolled off the finger tips.

But what does that matter in a world where agents are writing the code?

Certainly, you want some level of abstraction. I don’t want to be dealing with software that has a bunch of raw SQL statements in it duplicated throughout the codebase.

But do we need more ActiveModel abstractions? Does that hurt or help LLMs?

The folks at Remix are thinking about this as they build Remix v3: leaning into web standards while also making sure their API design choices pass muster of an LLM’s thought process.

So does it makes sense to continue iterating on Superflare?

Maybe. I don’t know yet.

Moving on

I’m not quite as pessimistic as Ken on this:

But I think there’s truth to what he’s saying: we’re living through a major shift in what is valued from us as software engineers.

My take: we lean into becoming builders instead. Learn how to operate agents and other machinery to provide value. Rather than obsess over code format and syntax, decide what techniques to leverage to create the world’s next big thing in the least amount of time.

I want to be like Adam, who’s doing a good job making lemons out of lemonade:

Or Jeffrey:

Or Aaron:

Let’s go.

Believe