June 19, 2026
Be in the Work, Not the Discourse
Here's a real one, pulled recently: "Microsoft AI CEO Has a Prediction for the Future of White-Collar Work Over the Next 18 Months." Most of what lawyers, accountants, project managers, and marketers do will be fully automated inside twelve to eighteen months. There's a new version of this story every week, and I can't help myself: I read most of them.
The pull to read them is strong. My first instinct is to disagree, strongly. And yet there's another pull underneath it, the read-on-and-see-if-it's-actually-true kind, the slow-down-at-the-car-crash kind. I keep reading even when I've decided the prediction is wrong.
And there's a separate urge entirely, one I feel more and more as these tools come online and more people pick them up. They are incredibly powerful. A lot of us have started using them to build things we'd carried around as latent ideas for years: products, software, the small thing we always wished a computer would just do. That pull isn't fear. It's the urge to make something.
White collar work has given me a tremendous amount. Material gains, a home, security and certainty. It also gave me an identity; for many years, being slightly corporate and slightly creative was who I was, rightly or wrongly. My work, and shamefully, my titles, gave me a language for myself, a way of seeing who I am that I'm not sure I'd have found elsewhere. So when these stories say this whole world might be ending, it feels personal.
About ten years ago I was on a flight, and a few rows up I could see a guy working in Salesforce. I had a distinct feeling: I could do this job. I want to say that humbly. He could probably do mine, and I couldn't have just dropped into his. But he was using a tool I knew, and using it in a way that mirrored things already in my own skill set. What I realized, sitting there, was that the work we were both spending most of our waking lives on was somewhat interchangeable, and the part we each regarded as deeply personal was, well, automatable.
* * *
I've realized none of this is fruitful. Not the stories, not the airplane theorizing, not the predictions, not the rationalizations about how this time is the same or different. The fruitful thing is to do the work.
Lately that's looked like a small, slightly funny session I host called Product Demo for Normals. A few of us building things with AI get on a call, and one person shares what they're actually making, live, screen shared, unpolished. This morning it was Stephen. He's building a product, and he walked us through a feature he was adding to it right there in the room, working with the AI while we watched.
What struck me wasn't the feature. It was how Stephen made decisions. At one point the model offered him two paths: plan the whole thing carefully first, or just build it and see. Early on in his project he did a lot of planning. He'd laid out the phases up front, built the structure to handle where it was eventually going. That planning was real and it was valuable. It's precisely because he did it that he can now move lightweight: put the thing on the screen, look at it, give feedback, ship. The planning earned him the right to stop planning.
Stephen isn't trying to win the argument about the future of building software, or the future of white collar jobs. He's building software. He's doing his job.
That's the whole intent of the room. It mirrors the one part of corporate life I'd actually want back: some accountability, some other voices, colleagues who'll comment on the thing. You show your actual work to a few people. You say out loud what you're making. The showing is the forcing function.
The work now is less about my identity than about making a specific thing. Whether or not I'm replaceable isn't mine to decide. The work isn't an argument, and speculating about where all this goes doesn't help. Doing the work does.