The clearest-eyed person I interviewed about AI this spring was a career coach in North Carolina, and she had used it to make herself replaceable. She works with graduate students on job placement, she is good at it, and she walked me through exactly how she built her own competition.

It started small. She is not a marketing person, so she handed her student outreach to a free AI tool and let it draft her emails. Then she went further. “I’ve already built bots with all of my brain inside of them,” she told me, “to do mock interviews, to do resume reviews.” She loaded in what she knows and gave them to her students, who loved them. The bots answer at midnight, she pointed out, which she does not. “I could build five bots, have them available 24/7, and I’m not sure anyone would need me, the human.”

She said all of this quite calmly, like it was arithmetic. “I already see the end of this job, because you can get hypercustom information out of AI.” She gives it her expertise; the student gives it their resume and the exact role they want. “The questions it asks are better than what I can come up with as a coach. As soon as I hand that over to the machine, the machine is better.” Then the part she has made her peace with: “I just don’t know how you justify paying my salary when you can have a free bot. And I’m using the free version.”

This is a forecast, not a hunch. She built the replacement and watched it get better than she is. She has even shortened her own estimate. A year ago she figured she had three years, and now she is not sure. “It gets better faster.”

The more AI does the work that happens on a screen, the more it is worth being a person someone can sit across from.

But she is not waiting around for it. When I asked what she would do next, she answered like a coach with herself as the client. Her bet is simple. The more AI does the work that happens on a screen, the more it is worth being a person someone can sit across from. So her plan is analog, on purpose. She wants to spend her days in a bookstore, what she called a third place, the kind we keep losing. “In a completely AI world, we still need something that’s tactile,” she said. She has even mapped the first step. 

If the job ends, she will take the severance and volunteer at the romance bookstore she already spends too much money in. Learn how it runs. Shelve stock, clean. Then make her pitch: a coach behind the counter, for anyone who walks in needing one. Volunteering is not the destination. It is how she plans to create a form of job security for herself.

Her calm should not be mistaken for a soft landing. She is a single mother with a son turning 12, no family money behind her, and a rehireability window she has already measured in years. 

“It’s middle-aged women that do the work that AI can quickly replace,” she told me. The worker who’s thinking the furthest ahead still has to improvise her own safety net. That is the part policy should not be calm about.

Logan Currie is a Visiting Fellow at Capita’s Family Policy Lab.