#OpenAI Yesterday, I published a research piece about finding a job using research. After reading it, the most resonant part for me wasn’t “Will AI replace jobs?”—it was that many people have already started quietly crossing into other roles to get work done.

The research sample comes from more than 800k job-related messages from US users, of which 16.8% involve cross-profession tasks. After stripping out general work like writing and summarizing, 43.5% of the profession-related messages are actually doing things that other positions would originally handle. Customer support starts touching data analysis; sales pulls up and looks at customer data themselves; marketing conveniently checks website issues. In the past, it wasn’t that many people didn’t want to do those things—it was that their skill set got stuck. They had to find people, queue up, and attend meetings. Now, people first throw it to AI and try it out. Many tasks that used to be outside someone’s personal competency can still be forced into completion.

This shift is even more obvious in small teams: in workspaces with 2–5 people, the share of cross-profession tasks is 18.9%, while in workspaces with 100 people or more it’s 16.3%. Small companies already lack headcount. Once AI arrives, what happens first isn’t layoffs—it’s one person starting to cover two or three roles at once.

I’ve been feeling this particularly strongly lately. In the past, I wanted to build a lot of automation, scripts, and tools, but I couldn’t write them, so it stayed stuck in ideas. Now the Codex and Claude quotas aren’t even enough, and I’m spending my days studying workflows and my nights continuing to tinker with automation. I almost want to treat one person like eighty donkeys. It sounds exaggerated—because it is—but productivity really has been pushed to a new height.

When I trade, I’ve never liked focusing on big phrases like “will the job disappear.” The market usually responds more precisely first: can one person carry half a role extra; can a team still charge the same way as before? The upside is efficiency goes up; the downside is also very real—boundaries become blurred, and the boss might assume you can do anything.

So my thinking is simple now: treat AI as an add-on first, and do as broad as possible with the work I can cover. As for whether I’ll be replaced—same as I said before—capital is always about cost-effectiveness!

#Codex #Claude