There's been a proliferation of "engineering" terms lately, kicked off by the infamous Prompt Engineer. Other new terms we've seen are Agentic Engineering, Loop Engineering, and Harness Engineering. I'm sure before the end of 2026 we'll get at least one more new Engineering term.
Whenever there's churn and thrashing, it signals that there's a point that's being missed.
All these fancy terms focus on the "how", whether that's asking the AI to do things in just the right way (for the prompt engineer) or building the correct type of box to put the AI into (for the harness engineer).
What is much more durable is to focus on the Why, and that's where the Product Engineer comes in. In a time where AI can make individual software engineers much more powerful, they need to take on greater ownership of what they're building.
Of course that's nothing new. To become a truly great software engineer, you always needed to care more widely about what you were building and why rather than just taking orders. It's just that, lacking that, you could always hide behind being really good at getting the computer to do things, even if you had no interest in understanding why.
Reflecting on my own contributions to the projects I've been on, the proud moments were never really about having written a particularly clever line of code and instead about having solved a real problem for real people, and that's where human judgment still provides a lot of value that an agent can't reproduce.
If you're a developer, I definitely encourage you to invest in your product skills. That's where I'll be putting a lot of my own focus for now.
