For a long time I could always find the edge of AI. The spot where it stopped being amazing and started being a party trick. Every time a new model dropped, I’d go looking for that line, and I’d find it. The models got better, the line moved, but it was always there. Honestly, I figured we were closing in on a ceiling.
I don’t think that anymore.
I used to be able to map the wall
Finding the limits used to be the easy part. Each new release, I’d push on it until it broke, and it always broke somewhere. It would lose the thread, make something up, or fall apart on anything that needed real depth. That failure point was reassuring. It told me where the tool ended and where I still had to take over.
I got good at drawing that line. I could tell you what AI was great at and exactly where it fell off the table. For a while, that boundary barely moved between versions, which is why I assumed we were near the top of the curve.
Then the line got hard to find
Somewhere in the last couple of months, the wall got faint. It’s still there in very specific spots, if you know where to push, but now you have to go looking. It stopped being obvious.
And the bigger change wasn’t the depth. It was the breadth. The number of things AI can be pointed at kept multiplying, faster than I could track. I don’t pretend to know even half of what’s out there. But the slice I do work with has been enough to change how I think about the whole thing.
One person can do what a team used to
Here’s where I’ve landed: I can’t think of a project or a business need that can’t be made better with the right use of AI. That’s a big statement, and I don’t make it lightly.
I’ve designed and built agents that create, reason, argue a position, chew through heavy calculations, and run entire processes start to finish. You still need a person in the loop, that part hasn’t changed and I don’t think it should. But the math is different now. One person with the right setup can do the work that used to take a whole team.
This is not the part where the jobs vanish
For the record, I don’t think this wipes out work. Done right, jobs evolve, the way they did with electricity, the car, and every other invention that felt like the end of something before it became the floor of something else.
Look at IKEA. When their chatbot, Billie, started handling routine customer service, they didn’t hand out pink slips. They retrained thousands of call-center workers into remote interior design advisors and stood up a whole new service line. Same people, better job, new revenue. That’s the good version, and it isn’t a fantasy. There will always be a bad version too. But done right, the upside is hard to overstate.
The part that gives me pause
All of this happened in about three years. That’s the piece I keep turning over.
Not the progress itself, which I’ve made my peace with and mostly enjoy. It’s the slope. If this is the distance we covered in three years, I genuinely don’t know where we’ll be standing after three more.
And for once, I’m not sure I can find the edge to tell you.