tl;dr AI let everyone do everyone else's job. PMs ship code, designers write specs, engineers make product calls. The panic that follows; what is my job now? - reads the moment wrong. Craft did not die. It got scarcer and worth more. What rises in its place is systems thinking: stepping back one click, questioning what you assume, and building your part so the whole works better. The quiet shift is not that everyone can prototype it is that everyone can now think early. The org that wins writes down what "good" looks like and bakes it into paved paths so more builders and more agents move fast without wrecking the place. And it hires differently: ask what tools people use, then let them use those tools in the room.
Give people the tools and product, design and data teams get much further before an engineer has to unblock them. A prototype, a first cut of code, a quick analysis are all reachable without waiting in a queue. Stone calls this a storming phase before a forming phase, and she is right: we are living in the mess before the order.
But speed with no target is just spaghetti at the wall; a thousand demos that answer no question anyone was asking. Point the new speed at a problem that matters, with your engineers in the loop, or you have bought motion, not progress.
If we can all build, do we still need separate functions?
We do. A PM still owns the what - is the problem even framed right? An engineer still owns the how - will it scale, is it sound, what will it break? A data scientist still owns the hardest question of all: can we trust this number? The early steps flow across roles now. The core craft does not. And one rule survives the blur: a human owns the outcome. An agent wrote the code. You still answer for it.
The "everyone can prototype" story gets the headlines. The quieter shifts matter more.
First, institutional memory becomes searchable. The old world locked hard-won knowledge inside the one person who sat through every experiment. Now you ask and get the answer in seconds - what we tested, when, what we learned - and you skip straight to is there something here I can use? You stop interrupting people to remember things for you.
Second, the early thinking goes democratic. It is no longer gated to the twenty-year veterans. Ops, finance, a country manager - anyone can now pull the threads, form a hypothesis and bring it back to work it deeper with a data scientist or an engineer. That is the step change. Not who gets to build. Who gets to think. That is the bigger deal.
Systems thinking is not a book with a slinky on the cover. It is seeing how the parts connect and building your part so the whole runs better.
Step out one click. Stone gives the cleanest version of this I have heard. For any task you are handed, stop and ask what larger problem it serves, whether your approach holds across the cases you have not seen yet, whether what you are about to build helps the people around you or only you. Then move. Do not camp in the question. A PM who questions forever ships nothing.
Think a level up. Do your job in a way that makes your manager's job and your colleagues' work easier, not just your own metric. Ask what your manager sees from their seat that you cannot see from yours - across product, ops and finance, not one team's number. Build so you leave a stronger version of the thing behind. Netflix has a phrase for the state this produces: highly aligned, loosely coupled. Agree on the priorities, then let people run.
Here is what that looks like on my ground. In PayGo device finance, the phone we sell is not a product with a lock feature bolted on. It is four things at once: the product (commercial terms, customer acquisition journey, onboarding flow), the collateral(intrinsic net value generated by owning the phone), the app that tells us how a customer behaves - attention data as a proxy of value, and the lock-screen switch that enforces the loan. A PM who treats the lock screen as a UI job has skipped the system; that screen is a risk control surface. Get it wrong and you have not shipped a worse screen - you have opened a hole in the credit book. The "how" decisions there; lock mechanics, the sign up flow, what a dealer can and cannot do cannot sit with engineering alone or product alone. They are owned together, or they are owned badly.
The market for narrow, deep specialists is thinning. The market for people who can move in several directions is growing. Concretely, that is: