Hey! Welcome back to another week of musings. We spent most of the weekend packing. So it was fairly busy!

I hope you had a more restful weekend and had time to rest.

Was this forwarded to you? You can subscribe here!

Things I enjoyed in the past week

• Stop being skeptical about AI for development with Charity MajorsCharity Majors on why 2026 marks AI's shift from hype to infrastructure, why code review might be overrated, and her rule to never send a message you haven't read yourself first.
• learning to savorJackie Luo on unclenching from rigid expectations, and how letting go of bracing for discomfort let her actually enjoy more of her life.

Before starting my weekend, the end of the week was stressful. We had an issue on a project that came out of left field. It was a lack of communication between teams, so preventable in hindsight, but hard to prevent during the project because everyone is doing their best.

Am I talking to you, or to your chatbot?

This post was triggered by another post I saw on Hacker News, "Don't be a meat proxy" and while the post feels harsh, it's something I've lived more and more recently.

I'll ask someone a question. Instead of an answer, I get a screenshot. I ask a follow-up, hoping for clarity, and I get another screenshot. If I ask about the hallucinated replies, the person doesn't actually know where it came from. It was like the conversation didn't happen.

I don't think this is really about laziness, though. I read something recently that made me reconsider the instinct to blame the person doing this. If someone reaches for a screenshot instead of an answer, maybe the real question is what's driving it. Are they overloaded? Is speed being rewarded over understanding, so the fastest path to "done" is to just forward what the model said?

That preventable incident I mentioned above, the one from a lack of communication between teams. I keep wondering if some version of this played a role. It goes to show that in large corporations, even if everyone is doing their best, they can still produce incorrect output.

I'm not sure there's a clear answer right now. People are tired or need to fast-track their tickets, and they need to get unblocked. And a lot of companies are tracking AI usage within their teams and individual usage, so the more you use it, the more you're seen as a poster child.

It incentivizes the wrong things. I've seen more and more "documentation" produced by AI, and people using AI to summarize it. On the flip side, people are asking teams for more and more documentation because they expect it to be AI-generated.

Your turn

Have you run into this, being handed a screenshot instead of an answer? How do you personally draw the line between using AI to move faster and using it to skip the parts of the job that require you to actually own the thinking?

Let me know your thoughts by replying to this email!

Happy coding!