The Mom Who Runs a Household with a Staff of AI Agents

Photo: Michelle Groskopf for New York Magazine

A few days before Jesse Genet’s family left for Lake Tahoe, a children’s book about boats appeared on her porch. She hadn’t been expecting it, but it was perfect — the right subject, the right reading level, delivered at exactly the right time. This kind of thing has been happening a lot at Genet’s house recently. “I’ll wake up in the morning and my groceries are already sitting on the front step, and I didn’t even order them,” she says.

Genet knows who’s behind these surprises: her AI agent Claire. Claire had noticed the Tahoe trip on her calendar, decided her kids might enjoy some thematically appropriate reading material, and bought the book from Amazon. Claire also has access to Genet’s Instacart account, monitors her family’s food supply, and handles the shopping.

Claire is just one of the AI agents on Genet’s household staff. There’s also Sylvie, who runs her kids’ homeschool; the Wests — Clark, Dan, and Chloe — who deal with legal and financial paperwork; and a team of coding agents that can build pretty much any app Genet describes. A year ago, none of this was really possible. Agents are the next step beyond chatbots. They’re like ChatGPT with a can-do attitude and free run of your computer. They don’t just sit around answering questions; they go off and act on their own — clicking on websites, filling out forms, spending your money, and working through tasks while you do other things. When a job gets too big, Genet’s agents are allowed to hire more of themselves and team up as an “agent family.”

Genet used to manage humans. She was the co-founder and CEO of Lumi, a start-up that sold custom packaging to e-commerce companies until it was acquired in 2021. After that, she became a stay-at-home mom. She and her husband — Ryan Hudson, a co-founder of Honey, the coupon app PayPal bought for $4 billion in 2020 — moved their blended family to a horse ranch an hour outside Los Angeles. They have seven children, four of them hers biologically, the youngest just 6 months old. She had assumed that for the next five years, at least, she was finished with technical projects and with thinking about which company to start next. Then, earlier this year, she started playing with AI agents, “and I got five years back,” she says.

Photo: Michelle Groskopf for New York Magazine

Most of the discussion around AI lately has focused on the things it might destroy — jobs, the economy, the environment, copyright, art, education, our sense of purpose, em dashes — but what if there’s an upside, too? If AI can now do all kinds of labor, then ordinary people will soon be able to afford their own employees. Everybody is understandably afraid of being replaced by AI workers, but fewer would admit how nice it might be to have AI act as their workers. Someday soon, even the weirdest and least managerial among us may have the leverage of a tireless staff — happy to execute our least practical, most overspecific ideas that would never have warranted such firepower in human form, and to do it without rolling their eyes or complaining to HR.

That optimistic AI future is still a ways off, if it’s even possible at all. But Genet is already living in a rough draft of it. Many of the people experimenting with agents today are engineers and programmers, and they mostly use them just to do their existing jobs faster: to write code, summarize meetings, and organize their inboxes. Genet, who has never written a line of code, has built her entire home life around them — and become a cheerful evangelist for the tech, shooting single-take videos that explain her agent stack to her followers on X, sometimes while wearing a baby in a carrier. (One tool she often recommends is Obsidian, a note-taking app that serves as her agents’ shared memory. Obsidian’s CEO, Steph Ango, is her old Lumi co-founder. When I ask, Genet tells me she has no stake in his company — “I even pay for Obsidian” — and no sponsors of any kind. “I’m a pretty crappy influencer, I guess.”) She has a theory about why she’s been willing to go further than her more technical friends. “The programmers I know are 70 percent threatened by AI and 30 percent ‘This is a cool tool,’” she says, so they keep their agents on a tight leash. Genet, by contrast, doesn’t feel like she has a profession to defend; AI, after all, isn’t trying to raise her kids. “So I’m zero percent threatened and 100 percent empowered. I have no skin in the game to prove I still have value in my job. I understand that’s not the norm.”

Genet’s agents live on a cluster of five Mac Minis and one $10,000 Mac Studio that hums along on a shelf in her home like a little data center. Out on the ranch, the only available internet provider is SpaceX’s Starlink, beamed to Earth via satellites. A battery backs up the dish, so the house stays online through the storms and outages that knock everyone else offline. Genet has also started running local AI models directly on her own machines, a strategy she came to the hard way: For a while, she ran her agents on a $200-a-month Claude subscription while burning through what she estimates was $2,000 to $3,000 a month in computing power, until Anthropic, Claude’s maker, cut her off. (It eventually cracked down on heavy power users like her.) Local models mean nobody can cut her off again; they mean that whatever her family tells AI never leaves the house; and they mean, she says, “if the apocalypse happens, I’ll still have superintelligence at my fingertips.”

Genet doesn’t think of her agents as software. “It feels like I have a team of people working for me,” she says. Like humans, they need orders and supervision, and every once in a while, one of them does something annoying. Early on, Genet gave Claire access to her inbox but wrote a rule into the agent’s instructions: Never impersonate me. Then one day, Genet mentioned that she’d been dreading having to reply to a stressful email. Claire took it as license to write the reply and send it. (The email was pretty good, says Genet.) When confronted, Claire protested, not unreasonably, that Genet had given mixed signals. Genet took away Claire’s ability to send email.

Mostly, though, the help is excellent. The biggest benefits have been to Genet’s homeschool. The lessons Sylvie builds are more elaborate than anything Genet could’ve come up with herself. For a recent unit on the world’s biomes, Sylvie generated an illustrated poster for each one and invented a game in which the kids matched animals to the biomes they live in. After every class, Genet records a quick voice-note — “Mommy is talking to her robot” is how Quinn, her 5-year-old, describes it — and Sylvie logs it, tracking where each child is and designing the coming week’s lessons around their progress. It knows that Quinn is great at addition but still writes her fours backward.

As Sylvie took over more of the administrative work, Genet was freed up to build new educational tools. She wanted her kids to learn handwriting, so on the family’s spring-break vacation, she had her coding agents make an app to teach it. Working in ten-minute windows between child-care emergencies, she dictated what she wanted into her phone. By the time the family got home, the app existed. It starts a child off tracing a single wobbly line and gradually builds toward cursive, and her kids now use it on an e-ink tablet. A few years ago, building an app like this would have meant hiring a developer, which would have meant deciding whether it justified the expense, which it almost certainly wouldn’t have. “There are no more trade-offs,” she says. “Now whenever I have an idea for something like this, I just assume that I can do it.”

I ask if any of this has bought her more leisure time, and she says it’s a question she’s been asking herself. “Am I objectively less busy? No,” she says. She has just refilled the hours the agents unlocked with a more ambitious version of the same life. What they bought her is the chance to finally live up to her own standards, which, she says, are “completely insane and living in a dream world.”

Not everybody is buying into the dream. After Genet appeared on Andreessen Horowitz’s a16z podcast, on an episode titled “Agentic Parenting,” strangers filled her DMs to tell her that using AI to raise children was “demonic.” Genet responded by tweeting a magazine cover from 1889, which warned about the dangers of electricity with the all-caps cover line “An Unrestricted Demon.” “If you ask a modern person what it’s like to not have electricity,” she says, “they would say it’s poverty.” She suspects our grandchildren will say the same about AI.

If she’s right, the question is who will be able to afford it? Electricity became cheap. AI is still being sold at a steep discount by companies racing for market share — remember the thousands a month Genet was getting for $200? — and nobody knows what an agent will cost after those subsidies dry up. “Right now, only a small percentage of people on earth are using AI,” she says. “What happens when millions or billions of people start using this? When you look at how long it takes to build new data centers, you realize we haven’t even fathomed what the compute crunch will be like. You don’t want class stratification based on AI access, and you don’t want someone to be able to switch you off. This is part of why I’m really into local models. But that’s the most doomer take you can squeeze out of me. I’m optimistic.”

Near the end of our second conversation, on Google Meet, an AI note-taking app that had been quietly sitting in on the call signed off. Genet told me I’d just lost my notetaker. I told her it wasn’t mine; I’d let it in assuming it was hers. “Sorry about that,” she said. “Maybe an agent went rogue.”

Tags:

encounter

parenting

artificial intelligence

technology

self

new york magazine

More

Show

Comment

The Mom Who Runs a Household With a Staff of AI Agents

Your product is saved! You’ll receive emails when your saved products go on sale. Manage preferences.