Gergely Orosz’s recent post got me thinking about something we tend to forget once we’ve been in tech for a while: learning is part of the job we’re paid to do.
It sounds obvious. But once you know your stack and can handle most of what lands in your backlog, it’s easy to get comfortable. Learning becomes something you’ll make time for after the release.
AI is making that comfortable spot harder to stay in. We need to learn new tools, but also reconsider where we put our effort. When a tool can produce a reasonable first implementation, more of the work becomes figuring out whether it solves the right problem and how it fits into the system.
You still need to understand what you’re shipping. In the 2026 Stack Overflow survey, 48% of respondents to the trust question said they trust AI when they can easily verify its output. Being able to do that verification takes knowledge. A convincing explanation from the tool won’t give you that knowledge automatically.
And there’s plenty of engineering work beyond the implementation. An unclear requirement needs sorting out. Another team needs enough context to make a decision. Something works in a test environment, but getting it into customers’ hands takes another round of coordination.
As a manager, I value engineers who help move that work forward. These skills take practice too, and I think they deserve more space in our development plans.
That’s what I’m preparing to talk about at conferences in 2027 with “Your Job Is Bigger Than Your Code: Engineering Impact in the AI Era.” I’ll share what I value in engineers’ everyday work and some practical ways to develop those capabilities alongside technical knowledge and AI skills.
Photo by Bluestonex on Unsplash.