I was on a weekend flight recently, half-watching the seatback and half-listening to Nikhyl Singhal on Lenny's Podcast. It's a good episode. He's been around long enough — Facebook, Google, lots of operator scars — to talk about how AI is rewiring our profession without sounding either evangelical or doom-laden. There was plenty in there I've been chewing on since.

But one observation has stayed with me more than the rest, and it has very little to do with product management.

He pointed out, almost in passing, that many women hit the peak of their careers at the same time they're navigating young families. And the implication of that, in a moment like this one, is uncomfortable: those women simply don't have the spare hours to immerse themselves in nights-and-weekends experimentation with tools like Claude Code. "We don't talk about it as an industry," he said, "but it's absolutely true."

I've been noticing it for months, but I hadn't quite named it. Once he did, I couldn't unsee it.

What I'm actually seeing

Building AI products is core to my job. I get to engage with these tools every working day. At home I tinker — small projects, some more polished than others. A weekend meal planner here, a script to tidy up our household admin there. Most of it never sees the light of day, but the muscle gets built. The intuition for what's possible. The reps that turn a tool from intimidating into something you reach for without thinking.

For many of my friends outside of tech, that's not the relationship they have with AI. It's much more transactional. Summarise this document. Reword that email. Speed up a thing that already exists. Useful, of course. But not the same as rethinking how you work.

That's not a criticism of them. It's a quietly logical response to a life that's already full. When you're senior enough at work that the stakes are high, and your kids are young enough that the evenings are short, there isn't a spare evening to spend playing with a new toy that may or may not pay off. Even when you know — abstractly — that you probably should.

Why this matters more than it looks

It's tempting to brush this off as a personal-time-management problem. I don't think that's what's happening.

The honest read is that the people who can put twenty hours a week of curiosity into AI right now are pulling ahead of the people who can't. And the people who can't are disproportionately the ones already juggling careers, partners, kids, and the invisible household load that still tends to land more heavily on women. The gap isn't about capability. It's about access to unstructured time.

If a meaningful portion of women are left behind in this transition, we risk undoing years of progress in representation at senior levels. That's not a soft aspiration. Diverse teams build better businesses — there's a well-evidenced body of work on that now, not a vibe. And the senior pipeline we've built across product, engineering and leadership over the last decade is fragile in ways we underestimate.

It would be a particular tragedy to lose that ground not to overt barriers, but to a quiet, structural one. Where the women didn't drop out, they just didn't get the same time to play.

What to actually do about it

I don't have a clean answer. But I have a couple of small things that I think are worth saying out loud.

First, for the time-poor women reading this: the bar for getting started is genuinely lower than it looks. You don't need to build an agent or learn to code. Pick one persistent friction in your week — a meal planner, a shared family schedule, a way to coordinate kids' activities, a system to manage school admin — and see if you can use AI to remove it. Start small, but start. The point isn't the output. It's that the muscle gets built on real problems you actually have, not toy projects.

Second, for the rest of us — leaders, partners, friends, organisations — there's a question I think we should all be sitting with: how do we create the space for time-poor people to build AI fluency, before the gap becomes structural?

Some of it is workplace design. Are we treating AI fluency as a thing people are expected to acquire on their own time, or as something we make space for during the working week? Are our internal experiments, hackathons, and learning sessions actually accessible to people who can't stay back until 7pm? Are we paying attention to who in our orgs is getting good at this and who isn't, and asking why?

Some of it is honesty about how the work at home is shared. The "extra time" myth tends to assume that one partner has nights and weekends free to experiment. For a lot of households, that's only true for one of them.

And some of it is just naming the thing. The reason Nikhyl's comment hit a nerve is that I'd been thinking about a version of it for months and didn't have language for it. Now I do. So do, hopefully, a few more people.

The bit I keep coming back to

I don't want, in five years, to be sitting in a senior team meeting that's quietly less diverse than it was in 2024, and tracing it back to this moment. The one where the technology changed faster than our systems for sharing access to it.

The good news is this moment is still early. The gap is real but it isn't yet structural. Six months of intentional effort — from individuals, partners, employers — can shift it meaningfully.

So pick the friction. Start small. And if you're experimenting, reach out. I'm always happy to compare notes.