I’m Not Exaggerating

I need to say that again because people hear it and assume I’m exaggerating or redefining terms. I’m not. I now ship more merged pull requests per week than any engineer on the team, despite spending most of my career managing the people who actually wrote the code.

And I’m not the only example. The person who ships the second most had never been a developer at all. He was an engineering manager. He ran teams. Now he consistently holds around 100 merged PRs a week.

That progression happened quickly:

“The Managers? They Go Ham.”

I made the prediction on November 10, 2025, before I had any proof. Three weeks later, I wrote, “The managers? They go ham.” At the time, one of those managers was still at the offer stage.

His first pull request landed December 15. It didn’t ship a feature or fix a bug. He added himself to the telemetry config so his output would show up on the board. Nobody asked him to do it. His first act on the job was to make the work countable.

That’s the manager instinct, and I think it explains a lot about what happened next.

Why Managers Won

The bottleneck isn’t typing code anymore. AI types the code. The bottleneck is knowing what to build, how to break it down, what good looks like, and when the work is actually done.

Engineers develop those skills over time, but managers spend most of their careers exercising them. Their job is already to take something ambiguous and make it executable — define the problem, scope the work, set expectations, review what comes back, catch what’s wrong, and decide when it’s ready.

Once execution becomes increasingly automated, those skills become much more valuable. The person directing the work doesn’t necessarily need to be the fastest person at writing code. They need enough judgment to give the machine the right work, in the right shape, and know whether what comes back is actually good.

That’s why I don’t think what happened here is magic, and it certainly isn’t because I suddenly became a great programmer. The robot is that good. A significant part of the execution that used to require years of technical practice can now be handled by the machine.

The industry is still hiring heavily for the part that got automated. What we’re seeing internally is that the skills surrounding that execution — judgment, decomposition, taste, context, and knowing what should exist in the first place — matter more than they used to.

Capacity, Not a Trophy

A hundred PRs a week isn’t the goal. It’s capacity. It tells you how much weight there is behind the arrow, not whether every problem needs that much force.

What changed once that capacity became possible was the range of things we could act on. Problems that had lived on a roadmap as “someday” started moving to “this week.” Small improvements that weren’t important enough to justify pulling an engineer away from something else suddenly became worth doing. Bigger ideas could be tested without committing weeks of engineering time before we knew whether they would work.

For me, that has been the biggest change. I can act on my own judgment, at speed, without waiting for someone else to build what I already know should exist. The PR count is simply the easiest way to see that change happening.

And the gap between the rest of the team and me has narrowed as the same capability spread. This isn’t about one person finding a trick and running away with it. It’s about making that capacity available to everyone.

An Experiment, Not a Study

This is still a sample of one company. It’s an experiment with early results, not a study, and I’m not claiming that every engineering manager is suddenly going to ship 100 PRs a week.

But the result is strong enough that it has changed what we look for in people. A traditional engineering background can still be valuable, but it isn’t the only signal anymore. Judgment, taste, the ability to define good work, and a willingness to actually use these tools matter a lot more to us than they did a year ago.

We have three seats open: security leader, engineering leader, and VP of Sales. All three assume this thesis. Title isn’t the filter.