
Over cocktails recently, a few colleagues and I kept circling the same idea: AI is making everyone a builder. Software engineering skill still gives you leverage, but you no longer need a formal engineering background to prototype, automate, and ship something useful.
Part of what feels different is how much cheaper the first draft has become. A rough internal tool, a workflow automation, or a small product idea no longer has to wait for someone with enough spare time to write everything from scratch. More people can test an idea while it is still cheap to change. That seems like a real change in who gets to participate.
We agreed that automating ourselves out of work is generally a good thing. If repetitive tasks can be delegated to tools, we should delegate them and move our effort toward harder problems, better judgment, and more original thinking. That does not mean implementation stops mattering. It means the bottleneck moves. The harder part becomes choosing the right problem, setting useful constraints, and deciding whether the output is actually any good.
That changes the center of gravity of work from execution to direction, from "how do I write this function?" to "what should we build, and why?" It can feel uncomfortable, especially in professions that used to treat execution skill as the main gate. I still think it is a net positive. Building is less restricted, capability is more widely distributed, and human taste, curiosity, and clarity matter more once the cost of producing something drops.
What stayed with me from that conversation was not the idea that everyone is suddenly an engineer. I already believed that more people were gaining access to the building layer of software. What felt useful was hearing that intuition echoed so clearly. That still seems good for experimentation, good for agency, and likely good for the range of ideas that get tested in the first place.
Get new posts by email
Subscribe for occasional updates when I publish something new.
Related posts
Medium and Meaning
February 22, 2026
A few reflections on building and creating with Codex and Sora.
From Writing Code to Orchestrating Agents
February 3, 2026
How my AI workflow changed from quick snippets to a practical plan/implement/verify loop.
Dropout as Implicit Bagging
March 7, 2026
Chapter 7 clarified that dropout works so well because it approximates bagging over many thinned networks with shared parameters.