There’s a moment, sitting at the kitchen table with my nine-year-old, when she’s working on her first real business. She’s trying to figure out what to charge for a four-pack of beignets. She’s done the math on flour and sugar and the foil bag and the labels. Her notebook is open. She looks up at me and goes, “Can I ask the AI helper to check this?”
And I say yes.
I want to tell you why.
I run companies for a living. I’ve built a renovation business and a coaching practice and a short-term rental portfolio and a real estate brokerage. I’ve been doing this for twenty years, and in the last two of those years I’ve watched something happen in real time that I think most people are still underestimating.
I’ve watched AI do, in fifteen minutes, work that used to take a whole marketing team a week. I’ve watched a single junior person, briefed well, produce a launch funnel that would have required three hires three years ago. I’ve watched the floor under whole job categories quietly drop two or three feet. The work didn’t disappear. The leverage changed.
Now look at my daughter. She is going to walk into the working world in nine years.
The kids who learn to think with this tool will compound, year over year, what they can do. The ones who are taught to fear it, or who never touch it because their parents are nervous, will spend the first decade of their careers playing catch-up. That’s not a take. That’s just what happens when a generational tool arrives and one cohort meets it on day one and another cohort meets it five years late.
So the question, as the parent of a builder kid, is not “should she encounter AI.” The question is “how, and with whom, and on what.”
Not a chatbot to befriend
Here’s what I’m not doing. I’m not handing her an open chatbot and walking out of the room. The danger people are right to worry about — the always-on, always-available, weirdly-eager-to-please thing that becomes a parasocial friend in a fifth-grader’s pocket — that’s a real danger. It’s the same danger as the scroll, the feed, the comment section. An app whose business model is keeping a kid’s attention is an app whose business model is opposed to the kid’s interests.
What I’m doing is different. It’s the difference between handing a child a saw and locking them in a workshop with a saw.
She uses AI in short bursts, on a real project, with me in the room.
She picks up the tool for a specific job. She does the job. She puts it down. The pitch she gives to her first customer is hers. The knock on the neighbor’s door is hers. The deciding-what-to-make is hers. The reading-the-customer’s-face is hers. The choosing-who-to-trust is hers. Most of the work of having a business, it turns out, is the human work — and that part stays human in our house.
But the math on her pricing? She talks it through with the AI helper, and then she decides. The flyer she’s going to print to put on the neighborhood corkboard? She writes the headline herself, then asks the AI to make ten variations, then she picks the one she likes and changes two words. Her business name? She brainstorms with the AI by saying, “I’m Rosie, I’m nine, I want to sell beignets at the neighborhood market. Help me name it.” She picks Texas Beignets because it’s hers and it sounds like her.
She is the founder. The AI is a helper. We say that out loud, every time.
What she’s actually learning
A lot of the noise about kids and AI assumes the AI is the lesson. Like the kid is going to grow up and need to know which button does what, what the model is called, what the API looks like.
She isn’t learning any of that. The models will be different in six years. The buttons will move. The interface will look completely unrecognizable.
She is learning something that won’t be unrecognizable: how to ask a good question, how to read an answer skeptically, how to keep what’s useful and throw the rest away. That’s the durable skill. The buttons are downstream of that skill, not the other way around.
She’s learning that when the AI gives her an answer that doesn’t sound right, she trusts her own brain over its brain. She’s learning to brief it like a contractor: “I’m doing X, for Y reason, and I want help with Z.” She’s learning that “Could you also…” is the most important sentence in working with a tool. She’s learning that the AI will make stuff up sometimes, and the only defense against that is her judgment.
I’m watching her become, in real time, the kind of person who will be able to use new tools as they arrive — instead of being intimidated by them. That’s the inheritance.
What stays human
Here’s the rule we write down: the AI never makes the thing. The kid makes the thing.
The AI never makes the ask. The kid makes the ask. We practiced the ask before her first market. Out loud. With me. “Hi, would you like to try some beignets? They’re two dollars for four.” No AI helps with that line. Nothing replaces a kid standing in front of a real person and saying it.
The AI doesn’t read the customer. The kid reads the customer. When somebody says “oh, that’s so nice, maybe later,” she has learned — and she learned this from being in front of real people, not from a screen — that maybe later almost always means no. The book calls that the signal. The signal is the human part.
The AI doesn’t sit with the no. The kid sits with the no. The AI is famously good at making a kid feel better. That’s actually a problem. Disappointment is data. A kid who has someone — or something — always available to soften every no is a kid who won’t develop the muscle for resilience. So when the third house in a row doesn’t buy, we don’t go to the AI. We talk about it. We notice what we noticed. And then we go knock on the fourth door.
In the book, I call this the sixth-door kid idea. The whole game changes at the sixth house, because you’ve stopped flinching by then. The AI doesn’t get her to the sixth door. She has to. I have to be the one who walks beside her until she does.
The parent’s job
The thing that makes this work, every single time, is that I’m in the room.
Not hovering. Not narrating. Not making her use the AI when she doesn’t need to. Just there. The same way I’m in the kitchen when she’s using the oven for the first time, and the same way I sit in the passenger seat the first dozen times she rides her bike on a real road.
When she’s about to ask the AI something, I notice the question. When she gets an answer, I notice if it’s the right kind of answer. When she’s about to take the wrong advice from the model, I gently say, “Hmm. Is that what you actually want to do?” and let her think. Most of the time she gets there on her own.
If I weren’t in the room, I’d be giving her a tool that’s eager to help her with anything, including the stuff I don’t want her getting help with. Being in the room is the whole safety mechanism. It’s why we built the book the way we built it. The book says, at every AI step, with a parent. Not with a friend. Not alone. With a parent. That’s load-bearing.
What I want for her
I want her to be a person who builds things. I want her to grow up able to spot a problem people care about, design something that solves it, ask for the sale without flinching, count the money honestly, and run it again next year better.
The world she’s walking into will reward the people who can do that work, plus use the new tools, plus know which parts of the work to keep human. That’s the combination. That’s what I’m trying to teach.
So she sits next to me at the kitchen table, and the notebook is open, and the AI helper is open, and the jar is on the counter waiting to fill up.
She asks her question. The model answers. She looks at me. She thinks for a second.
Then she decides.
That’s the lesson.
Jamey Ice is a dad and serial entrepreneur in Fort Worth, Texas. He wrote Kid Founder — The Young Entrepreneur’s Workbook for his own two daughters first. The free Founder’s Kit is at kid-founder.com.