Something a colleague said the other week stuck with me. She’d been at her company for about eight years and said the actual shape of her job had changed more in the last nine months than the previous six years combined. Not the title. Not the team. Just what she spent her Tuesdays doing.
That’s the version of the AI story that doesn’t really make headlines. There’s a lot of loud talk about robots, replacement, revolution, whatever. Meanwhile the actual shift is happening in the corners. Quiet stuff. Boring stuff, if we’re honest. And the tooling around it keeps expanding. Newer categories of ai tool are starting to do more than answer questions, they’re taking small actions on someone’s behalf, which is where the day-to-day starts to feel a bit different. Not dramatic. Just different.
For what it’s worth, a recent Pew survey found about 21% of U.S. workers now say some of their work involves AI. Which is a lot more than a year ago and, arguably, still less than the noise around it would suggest.
Anyway. Three places the shift keeps showing up.
1. The end of “I’ll just do it manually”
Every office job has that one task. The 40-minute reformatting job. The copy-paste-across-three-tabs ritual. The thing you don’t complain about because complaining takes longer than just doing it.
That layer is thinning out. Not vanishing. Thinning.
Some might argue this is the least interesting part of what AI can do, and that’s probably fair. But it’s also the part most people actually feel. A report that took two hours takes forty minutes. A spreadsheet clean-up that used to be a whole afternoon is now something you do while the coffee brews. Small wins. They stack up.
Side note, and this is a familiar theme, most of these gains don’t show up in any productivity report. They just disappear into slightly earlier finishes and slightly less-grumpy Wednesdays.
2. Meetings that (kind of) run themselves
Look. Meetings aren’t going away. That would be nice. Not happening.
But the admin around them? Different story. Auto-generated notes. Action items pulled straight out of the transcript. Follow-up emails half-drafted before you’ve closed the tab. Some of this is genuinely useful. Some of it is a bit wonky and needs a human to sort it out. Both things can be true.
There’s a fuller picture in this piece on how automation is changing workplace routines, which goes deeper than I want to here.
3. The trust question nobody’s fully answered
Here’s the one.
If a tool drafts your emails, updates your CRM, and books your calendar without you clicking anything, who’s on the hook when it messes up? Not in theory. In practice. On a Tuesday. When something’s gone wrong and the client is on the phone.
It seems that most organizations are still working this out as they go. There is reason to think that will change, but the pace is uneven. Frameworks like the NIST AI Risk Management Framework exist for exactly this reason, but let’s be real, nobody at a ten-person marketing agency is reading a federal framework on a Friday afternoon. They’re just trying to figure out if the intern’s shiny new automation is going to email a bad discount code to the whole list.
At least in theory, the governance stuff catches up. In the meantime, it’s a bit Wild West.
None of this is really settled. The shift is happening in the background of ordinary jobs, at ordinary desks, on ordinary Tuesdays, and it’s a bit hard to say where it lands.
Which is maybe the honest ending here. It’s changing. Slowly, then all at once, then slowly again.
