What Should Programmers Do While the AI Writes Code?

Hint: not start another AI coding loop

My cursor screen (even though I mostly use Claude now… but that’s a boring picture). Photo by me

Most every programmer is suffering the same growing pains right now. We’re Barry Allen (or Jay Garrick, if you’re old) with access to the Speed Force for the first time. Zooming around faster than our intent.

Like many of you, AI came crashing into my daily work-life. It paid dividends immediately. Things that used to take me days, I was doing in hours. I could take on large, complex, meaty projects and deliver them fast.

I was 5 times as productive. So, the next logical step? Well, I need to work on 5 times as many things, right? And that’s what I did. That’s what a lot of my friends and peers did.

A bug report that hints an infrastructure issue? I’m on it.

A new page for managing user settings? It’s in scope now!

An admin tool for block users? Put me in coach!

I had stacks of stacks of pull requests. AI was writing my linear tickets for me and responding to threads and writing project summaries. I was working on everything, everywhere, all at once.

But like a raccoon on head of a love-stricken chef, I wasn’t fully in control.

Raccacoonie from “Everything, Everywhere, All At Once”

And a few months in, I was burnt out. You want to talk about productivity? I managed to burn myself out 10 times faster than before! Now that’s speed for you!

I had to take a step back and ask myself: what should I do with this new productivity? Because clearly “more work but harder” wasn’t the answer. Context was getting lost. PRs would sit for weeks, forgotten. Reviews weren’t as thorough.

Slow work

Rather than continuing down that path, I took a step back and tried to think of the best ways I spend my time at work.

For me, the best times are when I’m learning something or helping someone else learn.

When my agent starts a loop, spinning up sub agents and self-reviewing and stacking PRs… what if… rather than starting another loop of work, I could just… read something?

What if I just DM’d a friend at work and asked if they wanted to pair on something?

Hell, what if I wrote something? (I know, as someone who works at the writing website, this one should have been obvious).

So that’s what I started doing. I started actually reading tech articles and blog posts. I’ve been in web development for over 15 years, and as such, I developed the skill of skimming the hell out of every article. I would search for the code snippet I needed, the conclusion, the link out to the library.

Before AI, you had to do that. Writing code took too long, so you couldn’t spare time digesting someone else’s rambling blog post.

And what you missed out on was the very human side of our industry. The funny writers, the shared life experience, the journey.

To be fair, most StackOverflow posts are worth skimming… but not all of them!

I’m reading documentation more thoroughly now. I’m understanding how a ViewTransition in React works, instead of just finding the code snippet, implementing it and debugging from there.

That thorough reading has informed how I talk to my agents. I am more consciously leading them down a path I would have paved myself because I’m taking the time to digest the full scope of a feature.

Writing

The best way I’ve been spending my time? Writing.

I write my daily journal on TK most every day. Before, I would have felt too busy to spare 15 minutes to write up some random thoughts.

And do you know what happened? People read them! And commented! I talked to real users and sparked ideas for new features. The submission rules we implemented? The content calendar? The new editors page? Those all came about because people read my work notes and left comments.

I can channel that productivity into something actually useful for real users. That’s worth the token cost.

I get the temptation. Your backlog never ends, your managers were wowed by the last thing you shipped in 1 day, so you want to chase that high, and everyone on LinkedIn is talking about the side project they’re launching.

But here’s what I learned by going slower and writing out my thoughts: careful work leads to more productivity in the long run. If you take the time think about your decisions, think about your product, write down your thoughts, or reach out to a coworker, you’d be amazed at what you can learn.

I’m not shipping 10 times as many things now, but the things I’m shipping are 10 times as meaningful.