There is a specific commit I remember. It was a Thursday afternoon, and I pushed a small fix—a validation check on an input field that had been bugging me for two days. The fix took forty minutes. The bug was real but minor. No customer had reported it.
That commit was the last one I made for three months. Not because I planned it that way. Because the next morning, everything else caught fire.
For the first two years of building, my calendar had one real job: write code, ship features, repeat. I was fast. I knew every corner of the product. When something broke at 2 a.m., I fixed it before the on-call Slack message finished threading.
That speed felt like value. It felt like proof I was doing my job.
But by year two, the business had outgrown what one builder could hold in their head. We had paying customers asking questions I hadn't thought about. A partnership conversation stalled because nobody followed up. A pricing change sat in a Google Doc for six weeks because I kept saying "I'll get to it after this sprint."
I was productive every single day. And the company was slowing down because of it.
The actual week wasn't dramatic. No blowup, no intervention. A co-founder sat across from me at lunch and said: "You're the bottleneck on three things that don't involve code."
She listed them. A customer expansion conversation. A vendor contract. A product decision that two engineers were blocked on because I hadn't written the spec.
I started to argue. She waited. Then she said: "You're choosing the work that feels good over the work that matters."
That sentence sat in my chest for the rest of the day.
Here is the thing nobody tells you about building a company: code gives you a feedback loop that nothing else in business does.
You write a function. You run the test. It passes or it fails. You know, within minutes, whether you did good work. The dopamine is small but reliable.
Compare that to the rest of founder work. You send a pricing proposal. Silence for four days. You have a product strategy conversation. The conclusion is "let's think about it more." You spend an hour on a hiring call and have no idea if the candidate is right.
Business work has long, noisy feedback loops. Code has short, clean ones. So when you are stressed, tired, or unsure, code is where you retreat. Not because it is the most important thing. Because it is the most legible thing.
I was not staying in the codebase because the company needed me there. I was staying because I needed to feel competent.
The first two weeks away from code were miserable. I would open my laptop in the morning and not know what to do first. My to-do list looked nothing like a sprint board. It was full of ambiguous items: "Figure out positioning for enterprise tier." "Decide if we need a sales hire." "Talk to three churned customers."
None of those tasks had a clear definition of done.
I kept opening the IDE out of habit. I would read a pull request and leave a comment, then catch myself spending forty-five minutes refactoring someone else's approach in my head. That was not code review. That was avoidance.
The honest truth: I felt useless for about three weeks. My output was invisible. I went to bed most nights unsure whether I had moved anything forward.
Around week four, things started to land. The customer expansion deal closed—because I had been in the conversation, not because I wrote a feature for it. We made the pricing change, and within two weeks we saw the effect on new signups. I wrote a product spec that unblocked the team for a full sprint.
None of that work was dramatic. All of it had been sitting there, waiting, while I fixed validation bugs nobody reported.
The company did not need me to stop coding forever. It needed me to stop treating code as my default. The ratio shifted. I went from ninety percent code to maybe twenty, and the other eighty percent turned out to be where the company was starving.
The hardest role change in a startup is not your first hire. It is not learning to delegate. It is accepting that the skill that got you here—the thing you are best at, the thing that makes you feel most like yourself—might be the thing you need to do least.
You will resist it. The resistance will feel rational. You will tell yourself the team is not ready, the code quality will slip, nobody else understands the system the way you do. Some of that might even be true for a while.
Do it anyway. The company does not need another engineer. It needs a founder doing founder work. And founder work, most days, does not compile.
Be the first to comment.
0 comments
Loading comments...