March 11, 2026
That's it, coding agents are just better than us at most well-defined, well-packaged coding tasks.
It is increasingly rare to find coding tasks where doing it manually will be more efficient than packaging the task clearly for a set of agents to tackle it. Case in point, we all felt the phase transition that happened between end of last year and beginning of this year.
This has important consequences on what I think are the three core drivers that make coding so attractive to me:
(i) They've removed the "hard tedious part" of our job that most of us were thriving for. I've always loved coding because it was hard. That feeling of accomplishment when all the parts come together and the system finally "works" after overcoming difficulty and tediousness. This was the craft, the style, the art of computer programming. All pretty much flushed down the toilets.
(ii) They've deprived us from the state of "flow" that many of us were thriving for. That unique experience where time flies, fully immersed in the task at hand. This state of "flow" was a big part of the appeal for coding to me. Coding with agents is an interruption management game. We're no longer burning CPU cycles on one powerful chip, we're instead pushed to emulating the most efficient task scheduler one layer up the stack.
But,
(iii) They also have dramatically amplified our ability to build and enact new systems in the world. Create something useful out of thin air by orchestrating the work of billions of transistors and thousands of kilometers of fiber. Making cool things happen with computers.
TBH, I've been a bit disoriented by this abrupt change of landscape. It's time for us to reckon with that new reality. I was irremediably attracted to coding in pursuit of the three drives above, only one is left.
Now that the cost of building is plummeting. The excitement remains "building" shit, and more of it. But the "hard" part has shifted. What has now become hard is, managing more context and incredibly faster context switching, and, even harder, deciding what to build with taste, technical taste as much as general taste. The fun part is that it's now possible to build many versions of a thing just to see what's best! The danger is bloating whatever it is that we build with too much stuff, forgetting to ship something pure and deliberate in the value it provides.
And as for the state of flow. Sorry to break it to you, no flow is the new flow.
We all need to look deep down at what was driving us as software engineers between (i) (ii) and (iii) and reckon with that new reality of where and how we can express our talent.