~/aakash
Writing ← Portfolio
Project Mgmt July 6, 2026

When AI Does the Satisfying Part, What's Left to Enjoy?

A junior developer I had trained told me the work had started to feel a bit hollow. I knew what he meant before he finished. This is a morale problem I did not see coming.

01

The hollow feeling nobody mentions

I spent eight and a half years as a backend developer before I moved into delivery, so I am not saying this from the outside. The fun was never the finished feature. It was the building itself, the quiet hours of making something work with your own hands.

So when a junior developer I had trained myself told me the work had started to feel a bit hollow, because the agent now does the part he used to love, I knew what he meant before he finished. I had watched this person learn to love building in the first place, and now the thing he learned is the thing the machine does. What is left for him is prompting and checking, and however fast that is, it is not the part that made him love the job. I have felt the same about my own old craft.

A team's morale is my job to watch, and this is a kind I did not see coming. It does not look like the usual problems. Nobody is overworked, nobody is stuck, the numbers are fine. Something quieter has just gone missing.

02

Faster is not the same as better

We talk a lot about what AI does to how much a team ships and almost nothing about what it does to how the work feels. Those are not the same thing. A tool can make a team faster and, at the same time, quietly take away the part that made the work good.

For a lot of people who build, the joy was never in the feature existing. It was in the making, in the flow of building something and watching it come to life. Give the making to a machine and you keep the result but lose the flow, and the result was never the part that fed anyone. You end up with a faster team that enjoys the work less, and those two things do not cancel out.

It is easy to brush this off as feelings. But it is not only about feelings, and the feelings would be reason enough anyway. It is about what keeps people doing good work for years instead of months.

A tool can make a team faster and, in the same motion, quietly remove the thing that made the work feel worth doing.

03

Morale was quietly holding things up

People who enjoy their work do it better, stay longer, and care more, and I rely on all three. A team that has become fast and joyless is not a stable team, even if the numbers look fine for a while.

The enjoyment was quietly holding things up, and taking it out has a cost that shows up late: people drift, switch off, or leave, long after the chart said everything was healthy. By the time it shows up as someone quitting, the cause is months in the past and easy to blame on something else.

That delay is the dangerous part. A morale cost you feel today, you notice and fix. A morale cost that turns up two quarters later, as your best people slowly losing their spark, you often do not trace back to what caused it. So it never gets dealt with, because nothing on a dashboard is flashing red.

There is a quieter effect on the people who stay, too. When the satisfying work goes to the machine and the checking stays with the humans, the day slowly fills up with the parts nobody chose this job for. Even people who do not leave can end up in a version of the work they would never have signed up for, and that is hard to spot until it has already happened.

04

Drawing the line on purpose

I do not have a neat fix, and I do not trust anyone who says they do. But the honest thing is to at least name it, instead of pretending the trade is not real. Some of the enjoyment can move rather than vanish. A few people find real pleasure in shaping what gets built and why, and for them this is a better job, not a worse one.

But it would be dishonest to pretend that is everyone. Plenty of people became developers because they love building, and telling them to enjoy checking a machine's output instead is a bit like telling someone who loves cooking to be happy inspecting meals somebody else made.

So what I try to do, imperfectly, is protect some of the building where it still makes sense. Not everything has to go to the agent just because it can. There is a real case, one I would have called soft a year ago, for letting people build some things by hand because it keeps them engaged, even when the machine would be quicker. A slightly slower team that still enjoys the work may well beat a faster one quietly running out of reasons to care.

I am not sure exactly where that line sits, between protecting the craft and just indulging it, and I do not trust the confident answers either way. What I am fairly sure of is that it is a line worth drawing on purpose, instead of letting speed draw it for you. Handing everything to the machine because you can is a choice too, even if nobody says it out loud.

So I will ask the thing I have not fully worked out: when the machine does the satisfying part, what are you doing to keep the work feeling worth doing, for the people who used to love it most?