A field report from a Config Leadership Collective session on building AI-native orgs, and why roles perforate rather than disappear.
Last week I moderated a table at Figma’s Config Leadership Collective, in a session built around one question: how do you actually build an AI-native organization? The room ran on Chatham House rules, so I’ll keep the names out and credit the ideas to the roles they came from. The thinking was good enough that I want to pass it along, with my own reflections layered on top.
The line I keep repeating came from a product leader at an AI-native legal startup. She said the boundaries between disciplines aren’t disappearing, they’re perforating. Everyone can reach into design, engineering, and product in ways they couldn’t a year ago, and that access is real and worth celebrating. But the holes are perforated for a reason. When a build goes off the rails, it’s usually because the team tried to make the holes vanish entirely, and suddenly nobody owns the recommendation because everybody does. Her fix was simple: one person still has to go deep, validate with customers, and hold the call on prioritization. Perforated, not erased.
Her colleague on the panel, a design leader at the same company, had the sharpest take on prototyping. When anyone at the table can spin up a coded prototype, you stop arguing about whether you all read the doc the same way. You put ten prototypes on the table and look. Ten that match means alignment is real and you can see it. Ten that diverge means it’s time for the conversation you were avoiding. The prototype becomes the alignment instrument, and product definition that used to take weeks now takes an afternoon.
That same speed has a tax, and she framed it quite well.
Teams with high ownership will run the second you say run. The trouble is they’ll run without asking where.
Her anecdote stuck with me: you say go, the team goes, and a few days later you realize nobody asked which direction. So the leader’s job shifts toward something less glamorous than velocity. It’s manufacturing clarity faster than the team can sprint past it, recalibrating what good looks like on a near-weekly basis, and being honest that you’re figuring it out at the same time they are. She had a phrase for the hiring side of this too: the framing moves from ‘I’ll scale you’ to ‘you’re going to unlock me.’ I’ve been thinking back on that line all week.
On hiring, the room landed somewhere I agreed with. Pedigree is a weaker signal now. What you’re looking for is closer to a cowboy tinker, someone goal-oriented who also loves to poke at things, because curiosity stopped being a differentiator and became table stakes. The portfolio tell is the same one I’ve used for years: all process, no outcome. Fifty slides of method and one timid ‘where did it land’ slide tells you the person hasn’t owned a result. What’s new is the last quality, and it’s the one I closed my own table on. It isn’t just the willingness to learn, it’s the courage to challenge what you’re learning, to push back on what an agent hands you instead of taking it at face value. The better we teach these tools, the better the leaders we grow around them.
My own table spent its time on two threads worth adding. The first is that brand and product design are converging in a way I’m enjoying. Our marketing designers used to make simplified, abstract versions of our UI for the marketing site, and now those simplified versions are informing the actual product, with the two crafts starting to build in the same file instead of exporting PNGs at each other. The second came from a founder who’s the combined product, design, and engineering lead at a five-year-old company. Single ownership solved his silo problem completely, everything ends with him, until it didn’t, because somewhere past 150 people one person becomes the bottleneck. The room more or less agreed that as the lines perforate you need more decision-makers at lower levels, not fewer, which means getting flatter and a lot more deliberate about who owns what.
The through-line, if there is one, is that none of this removes the need for judgment. It relocates it. When everyone can build, building the right thing is the hard part, and that is squarely a leadership job.
The teachable part
When everyone can build, judgment doesn’t get cheaper, it gets scarcer. The leader’s job is to manufacture clarity and make sure one person still owns the call.