Women picked up ChatGPT suspiciously fast, not because we suddenly fell in love with technology but because we've been writing prompts our whole lives.
We explain a task to a boyfriend in a way that won't offend him, to a child in a way he’ll understand, to a boss in a way that makes him think it was his idea, and to a mother-in-law in a way that lets us simply survive. A prompt engineering course costs a thousand dollars. Some of us got the skill for free along with the second X chromosome.
A good prompt is just a normal conversation, except the other party finally doesn't roll their eyes or say "here we go again." It involves explaining the context, calibrating the tone, noticing when a point it didn't land, and trying again from another angle. For decades these abilities were called soft skills and treated as something sweet, feminine, secondary — compared to real, hard, masculine expertise. Then LLMs arrived, and it turned out that soft skills are exactly what makes working with them effective.
Here's what's happening to me, a grown woman with twenty-five years of experience in digital careers, survivor of a couple of tech revolutions. I look at LLMs roughly the same way you look at a new hire on day one. Smart, you say? Well, let's see.
We didn't become prompt engineers — we always were. We’ve developed those skills as executives, CMOs, mothers, wives, daughters of aging parents, career women and philanthropists — usually all at once, because how else would any of these jobs get done? And with LLMs, it seems we're doing better than the people who built them.
Chapter 1. Morning: AI as a Second Cup of Coffee (Sometimes the First)
My mornings used to start with two sources of anxiety: my phone and my conscience. The phone showed me what I didn't finish yesterday. My conscience added that I wouldn't finish those things today, either.
Now there's a third entity wedged in between. I open Claude with roughly the same frequency my mother opens the fridge, and for the same (lack of) reason: just to see what's in there and whether it might come in handy. It almost always does.
Eight a.m. means coffee. Everyone's still asleep. In the twenty minutes before the official start of the day, I get through four things. I nudge the building manager of the Milan apartment, check Claude to find out how Dad's new medications interact with his old ones and forward the notes to Mom, phrase a reply to partners so our "no" sounds almost like a "yes," and sketch out talking points for the board. The coffee is still hot.
This isn't "using artificial intelligence." This is everyday magic. You get used to it in a week and then can't remember how you ever lived without it.
My friends tell the same story, just with different scenery. A gallerist in Milan decodes her dreams with ChatGPT — and now, it seems, she dreams more than she did in her entire life, or maybe she's just finally remembering the ones she has. An Iyengar yoga teacher in Tel Aviv builds individual programs around each student's injuries, limits, and progress. Every morning I drill myself on Italian verb tenses. Gemini writes the exercises in thirty seconds.
It's precisely in these very personal, very specific tasks that language models have become indispensable to us. AI gives back what's always in short supply: time and attention. It helps rebook a connection, draft a press release, and write a letter in a foreign language so that it sounds human, not corporate — a task that used to take forty minutes. Having that time back lets me take my time talking to Mom, rather than squeezing the call in between everything else.
A small revolution is unfolding in my kitchen, at the corner store, and in the line at the visa center: all the places where revolutions don’t usually happen, or where nobody notices them.
Chapter 2. Why We're Better At This
I don't write prompts.
I talk to a model like I'd talk to a person. Keeping in mind who I’m talking to (Claude or Grok), I explain specifically what I need and give examples. If the LLM doesn't get it, I try a different angle. If that doesn't work, I rephrase.
This never seemed like a process I needed to explain. And then it hit me.
What I'd always thought of as "just having a normal conversation" turned out to be an excellent professional skill. I didn't even have to learn it on purpose (because girls are expected to come with it preinstalled, right next to tying your shoes and brewing tea).
I can calmly walk my grandmother through WhatsApp without reducing her to tears, or gently tell my boss that his idea is less than brilliant. I can even talk my boyfriend into something he absolutely doesn't want to do — and make him think it was his idea all along. (I'm especially good at this one. Honey, if you're reading this: sorry, now you know the terrible truth.)
All of this goes by one boring term: emotional labor.
The sociologist Arlie Hochschild coined it while studying flight attendants. It’s the art of smiling when you don't feel like it; calming someone down while your own nerves are shot; and reading and adjusting to every passenger's mood. Every woman on the planet works as one of those "flight attendants" from time to time — just on different routes.
When LLMs entered our lives, it turned out that our emotional labor maps almost perfectly onto what's called prompt engineering. The language model just joined the lineup with the neurotic boss and the cranky toddler. "How do I phrase this so he won't take it the wrong way and will actually get it?" becomes "How do I phrase this so she'll understand and produce the right output?" Same logic.
See for yourself what prompt engineering courses and clever books recommend:
"Provide context"
Any girl calling her mother for advice does this automatically. Without a twenty-minute preamble, Mom won't understand a thing, but she'll start panicking immediately.
"Specify the role and tone"
Tell the model who it is: expert, friend, critic. Any woman who's ever asked a friend to "just listen, don't give advice" knows how critical that instruction is. Without it, the friend starts trying to fix things (when usually there's nothing to fix; you just need someone there).
"Give examples"
Show what "good" looks like. Any mother knows that “clean your room,” without an example, can mean “pick two socks up off the floor” or "deep-clean the place and reorganize every closet."
"Iterate"
Didn't work the first time? Rephrase, restate the task, try a different angle, say it again. Any woman who's spent years explaining to a boyfriend or husband why you have to book flights and vacation time in advance has Olympic-level mastery of this skill.
A look at my chat history and those of my friends shows emails, talking-to-myself drafts and talking points for difficult phone calls. My boyfriend’s chats and those of his colleagues contain code, regex, debug logs, and SQL queries. This isn’t a representative sample, but it is a consistent one.
Some will say "women use AI superficially," but I'd put it differently: women use AI exactly where it actually changes daily life. It doesn't just save us time — it saves emotional energy, the most delicate and expensive resource a working woman has, and one that she burns through faster than her morning coffee.
When someone tells you, with that little condescending lilt, "oh, you just asked ChatGPT," you switch on the flight-attendant smile and think: darling, I've spent twenty-five years learning to phrase my requests so they actually get done. And honestly, I get better results from the models than I do from people.
Chapter 3. No Longer Doing It All Alone
For twenty years I ran a digital agency with a hundred people on the team. That's not a number in an HR report — that's a hundred lives passing through my office, my inbox, my weekends, and my sleepless nights. I knew everything about every one of them: whose baby had just been born, whose mother was ill, who was prepping for surgery. They all needed advice and support because we were a team, which is almost a family.
Alongside all that came the work, the clients, the projects, and the plan the shareholders had approved. That plan focused on growth, margins, attracting new clients, and retaining old ones. It didn't care that one of your key people was going through an ugly divorce or that someone else had just lost their father.
I lived between the work reality and the human one, learning to hold them together without splitting in half. If you don't remember who's hurting and where, you won't have a team. If that's all you remember, and you forget the plan, you won't have a business.
Holding empathy, strategy, and Excel in your head all at once is a skill male executives generally don't have. That’s not because they're dumber or less empathetic by nature. It's just that there's always someone next to them who remembers the birthdays, books the meeting room, and keeps track of who on the team is having trouble at home. This invisible someone is, as a rule, a woman: a wife, an assistant, an HR director. Any structure that lets you "focus on strategy" is based on somebody's invisible labor.
Now, for the first time, I have a real assistant. Not the kind you spend more energy managing than it would take to do tasks yourself, and not in the “write this email for me” sense (although, sure, that too). I have someone with whom I can think through problems, like a conversation with an employee I have to fire or a board meeting where I have to explain why we missed our margin and not look weak doing it.
I used to have these conversations with myself. Or I’d reach out to a friend, who has her own three companies, forty meetings a month, and a mother recovering from surgery. Now I talk the situation through with the model, hear three possible framings, pick the fourth — the one that came to me while I was reading the first three — and walk into the meeting with a clear position. An hour of internal monologue becomes ten minutes of productive conversation.
No, it's not a substitute for human advice. I still call my friend and my parents, still talk things over with my boyfriend whenever his opinion is the one that actually matters. But that whole massive layer of work I used to run through my head at night — I can finally put it outside myself. On the screen. In the chat. And finally get some sleep.
Chapter 4. No Time to Design the Future
The obvious conclusion from what I’ve written would be: "women are naturally strong at working with AI, so the future belongs to us." This reads beautifully as a LinkedIn post. It’s a shame it's not true.
History has pulled this trick before. For centuries, textile production rested in women's hands: spinning and weaving were mass female labor, often the only source of independent income. Then came the factories — and women didn't go anywhere. They just stopped being craftswomen and became cheap labor instead, with no voice and no share of the profits. Status dropped, pay dropped, independence vanished.
Another example is programming. In 1843, Ada Lovelace, daughter of Lord Byron, published the first computer algorithm in history for Babbage's analytical engine. She's still considered the world's first programmer. And there's more: ENIAC was programmed in 1945 by six women, Grace Hopper built the first compiler in 1952, and for decades NASA's calculations were done by "human computers," who were also women. Then the profession became prestigious and lucrative — and women somehow vanished from it.
That's a reason not to get comfortable.
The rules of the game with LLMs are being written right now. On the teams designing the models and deciding how they'll work, women are still a minority. Among the authors of key AI publications, they make up only about 16%.
This isn't happening because we're being shut out. It's because our hands are full: teams, kids, quarterly plans, someone running a fever of 100.8. There's no energy left for designing the future.
That's the real trap. We're becoming the best users of a technology built without us, playing virtuoso on an instrument shaped for someone else's hand.
I'm not telling everyone to go work in AI, but if you have a voice, use it. Ask the uncomfortable question at the conference.
And write what a man won't write.
