AI does not spread through a company on its own.

I was reminded of that watching Jiaona Zhang. She runs product at Laurel now, and before that she led product at Airbnb, Webflow, and WeWork. She sat down with Aakash Gupta, screen-shared the “Company OS” she built in Claude Code, and walked through how she actually rebuilt the way her company operates. I’ve watched a lot of AI content this year. Most of it is people selling you slides. This was someone showing you the repo.

A few things stuck. One of them I’d push back on. I’ve been turning it over since, so I’m writing it down.

The gap she named better than anyone

Every company, she says, has a 1% who are already AI-native. The tinkerers. Their workflows are locked in. Then there’s “the 90 to 99% of the rest of the organization who isn’t sure what to use when.” And the value the 1% creates, in her words, “compounds in a bucket” nobody else can reach.

I’ve watched exactly this happen from the inside. A leader buys the seats, sends the memo, waits for the productivity. One or two people take off. Everyone else opens the tab twice and goes back to how they worked before. That is not a discipline problem. It is the most predictable thing in the building.

I run my own fleet of agents, and even for me the gap is real. The stuff I’ve figured out lives in my head and on my machine. If I don’t deliberately turn it into something someone else can pick up and run, it doesn’t travel. It just sits with me, getting sharper, helping no one else.

So when she says the fix is to encode that 1%‘s workflows into skills anyone can call, I nodded. That is the actual problem. Not which model you bought. Whether the knowledge moves.

Someone has to own it

Her line on this is the one I keep repeating to leaders. “A lot of times when you say it’s everyone’s responsibility, it’s no one’s responsibility.”

Her fix was to give AI operations to one full-time person whose whole job was to demonstrate value. Once that person visibly pulled ahead, every function wanted their own. She calls it the new biz ops. I think that’s right, and I think most companies get it exactly backwards. They make AI a line in everybody’s job description, which is the surest way to make it nobody’s.

She’s also honest about the trap on the other side. She over-automated her own tasks, scheduled so many jobs she hit information overload, and had to admit you “can’t just assume that people are going to do this for themselves, and second of all, that they’re not going to be overwhelmed by the number of automations.” I liked that she said it out loud. More automation is not the scoreboard. I’ve seen people automate themselves into a mess they now have to manage.

Her maturity ladder is a clean way to see where you actually stand: level one is chatting with the model, level two is automating a single workflow, level three is building a small app, level four is building apps other people share. Most companies I talk to are living at level one and telling themselves a level-three story. Naming that honestly is worth more than any tool purchase.

Where I’d push

Now the part I’d push on. Not because she’s wrong. Because I think there’s a whole layer underneath her framing that decides whether any of this works.

She talks about the org design, the skills, the ontology, the captain model. All of it is good. What she mostly leaves out is the fear.

Underneath every AI rollout is a quiet question your people are carrying. Will I still have value? Brené Brown has named fear of irrelevance as one of the dominant things people feel at work during change. I’ve seen it stall rollouts that had perfect tooling. And the part leaders rarely admit is that they’re carrying the same question while they convince everyone else of the answer. You can build the cleanest Company OS in the world. If your people think it’s the thing replacing them, they will quietly route around it.

Which leads to the bigger place I’d extend her. The instinct is to treat people as the cost AI takes out. I’d argue the opposite, and I’d argue it as the actual mechanism. Your people aren’t the trade-off. They’re how the value shows up at all.

There’s data pointing the same way. In BetterUp’s Uplift 2026 keynote, CEO Alexi Robichaux shared research that is uncomfortable for the tool-first crowd: AI adoption on its own does not reliably move performance. What moves it is what the manager does with the time AI gives back. Managers save about six hours a week with AI, and only about one in eight or nine reinvest it into developing their people. The ones who do drove a 65% increase in their team’s AI performance. The way to raise your team’s AI adoption was not more tools or more training on tools. It was pouring the saved time back into the people. The coaching around the seats mattered more than the seats themselves.

That’s also where I’d re-read her best story. She went from leading hundreds of people to five PMs and four designers, and the team ships more than it ever did. The reflex is to hear “smaller team wins.” I read it differently, and I think she would too. The lesson I take is about where you point the people you have. Automate the tedium so a person spends their day on the work only a person can do, the judgment, the customer, the hard call about what to build and why. She said it herself about her own field: the fundamentals matter more than ever now, only the speed changed.

So the prize was never a leaner headcount. It’s people freed up for the part of the work that’s actually theirs.

What I keep coming back to

The through-line for me was simple. All of it only pays off if the humans in the building believe they are being invested in, not phased out.

This is the conversation I keep having with leaders. Someone owns the gap between your 1% and everyone else, or no one does. And somebody is either coaching your people through the fear, or letting it quietly eat the rollout.

If that’s the conversation you’re in right now, I’m happy to sit in it with you. Book a call and we’ll talk through where your company actually stands and what to hand off first. No deck. Worst case, you leave with a clearer map than you came in with.

So, in your company: who owns it?

Sources

  • Jiaona Zhang, Chief Product Officer, Laurel. Interview on the Product Growth channel (Aakash Gupta), “How to Build a Company OS in Claude Code,” 2026-06-24. Video · Transcript writeup.
  • BetterUp, Uplift 2026 keynote by co-founder and CEO Alexi Robichaux, “The Value of Human Work in the Age of AI.” Keynote video · session highlights.
  • Brené Brown on fear of irrelevance as a dominant feeling at work during change.