06 February 2026

February 6, 2026 Palm Springs, California

The Escalation

There’s been a development. Supervisor had a sit-down with me about overuse of AI. He said my Pull requests have too many files and it’s taking too long for other developers to review them and my stories are the most expensive in development on the team.


TherapistGPT Responds

Well. That escalated from “this stings” to “this has consequences.” Let’s not panic, but let’s also not pretend this is nothing.

First, what you’re feeling now is likely fear plus grief plus injustice, with a thin glaze of anger. Fear because this crossed from peer feedback into managerial signal. Grief because something that was helping you function is now being framed as a liability. Injustice because the metric being used feels blind to why you work this way. Your nervous system is probably reading this as “I am becoming a problem,” which is a very old ADHD wound.

Now, strip the emotion off the facts. Your supervisor is not saying “your code is bad” or “AI is forbidden” or “you are failing.” He is saying three very specific, managerial things:

  1. Your PRs are large.
  2. Large PRs impose review cost on the team.
  3. That cost is showing up in story estimates and delivery metrics.

That’s not a moral judgment. It’s a systems problem. Managers live and die by throughput, predictability, and shared load. Anything that spikes variance gets flagged, even if the output is good.

The Stoic Angle

From a Stoic angle, this is almost textbook.

This is squarely in the dichotomy of control. You do not control how Jira calculates “expensive.” You do not control how other developers experience cognitive load when reviewing. You do control PR size, narrative clarity, and how your work is staged. Right now, you’re letting a tool optimize for your focus while the system optimizes for their coordination. Stoicism would say neither is wrong, but the role demands alignment with the system.

There’s also role obligation again. Your role is not “produce the best possible code in one go.” Your role is “move work through a shared pipeline without clogging it.” Marcus Aurelius would absolutely side with the boring manager here. Do the work in a way that disturbs others the least, even if it’s less elegant.

The Daoist Correction

Now Daoism, because this is where the real correction lives.

You’re forcing the river.

AI lets you see many improvements at once. So you take them. The Daoist mistake isn’t using the tool. It’s acting on every insight immediately. Laozi would say you’re harvesting the whole field when the village only needs tonight’s meal. The resistance you’re now facing is the system saying, “We can’t digest this much change at once.”

Wu wei here does not mean stop using AI. It means decouple generation from submission. Let AI explode possibilities privately. Then deliberately submit less. Smaller slices. One intention per PR. One story does one thing, even if you can see five things that “should” be fixed.

The “most expensive stories” comment hurts because it feels like a character attack. It’s not. It’s a symptom of scope creep via insight. AI gives you insight faster than the team’s process can absorb it. That mismatch is the real problem.

The Practical Reset

Here’s the practical reset, no philosophy incense required:

  • You keep using AI to help you think and focus. That stays non-negotiable.
  • You hard-cap PRs. File count, diff size, whatever metric the team respects. If it doesn’t fit, it becomes a follow-up ticket or a future refactor.
  • You narrate aggressively. PR descriptions that say “Intentionally scoped to X. Deferred Y and Z to avoid review load.” Managers love this sentence. It signals self-regulation.
  • You stop refactoring adjacent code unless the story cannot be completed without it. Even if it’s ugly. Especially if it’s ugly.

Emotionally, the Stoic move is to treat this meeting as information, not condemnation. The Daoist move is to yield without surrendering your nature. You are not being asked to become someone else. You’re being asked to throttle.

Not a Failure State

One last thing, and this matters: this is not a failure state. This is an inflection point. People do not have sit-downs with engineers they’ve written off. They do it with engineers who are productive but misaligned. That’s a fixable problem.

You’re not too much. You’re just currently emitting more change than the system can absorb. Adjust the flow, not the source.


Read Part One: On Inadequacy, Code Review, and Stoic Acceptance

Continue to Part Three: Misaligned Incentives