Build the Mad Thing

In Greece the death notices go up in public. Taped to a lamppost, an electricity pole, sometimes a tree. It is an old tradition. Once you start reading them you cannot stop looking at the ages. Ninety one. Ninety four. Almost never a number that makes you flinch.

One in three people here reach their nineties. There is no secret to it. The food is real, the days unhurried. You live your whole life among people who need you, on an island too small to opt out of each other. For most of us, none of that is the default any more. It is a decision.

And the decision is getting heavier, because the years are coming. Medical AI will raise the floor under a human lifespan the way electricity raised the floor under a working day. More of us will get the extra decades the island already has. Which turns a quiet question into an urgent one. Not how many years. What you put behind the number.

There is a strange rhyme in that, because a floor just rose under something else too. We are all holding the same magic wand.

An app that once took a year to build gets scaffolded in an afternoon. A plausible essay, a plausible strategy, a plausible coaching conversation, all in seconds. The barrier to making things has collapsed, and the result is arriving on schedule. An ocean of output that all looks and sounds the same.

None of this is broken. It is the maths of the tool. Millions of people prompting the same models toward what already exists produces convergence. AI raised the floor a long way. It raised the ceiling too, but nowhere near as far. When the floor climbs faster than the ceiling, the middle fills up. And the middle is where nothing distinctive lives.

So the question is not what can I make faster. It is what can only I make at all.

There is a second shift under the first. It is not only what you can make that changed. The rules for building anything of value got rewritten too. Purpose, passion and product used to sit far apart. You reached one by grinding through years of the others, through capital and headcount and permission, and most people spent a whole career never closing the gap. You built the fundable thing, not the one you meant to build. That distance has mostly gone. There has never been so little space between what you care about, what you are good at, and what you can actually ship.

Put fifty more years of living next to that, and the question changes shape again. The safe answer stops being interesting. What comes next?

The floor rose. Your ceiling did not move to meet it.

My grandfather's generation needed something close to a chemistry degree to develop a photograph. Then digital cameras arrived and anyone could take a thousand pictures a day. Did photography die? No. Most of those pictures sit forgotten in cloud drives. A few still stop you mid scroll. The making got commoditised. Taste and timing did not.

We all have the same English language. We do not all write equally well. Equal access to a powerful tool has never once produced equal results. AI will not be the first exception.

Why it feels like theft

AI never threatened anyone's skill. It threatened the I am.

We learn a craft, work at it for years, and quietly become it. I am an engineer. That is the trap, in four words. When the doing gets absorbed by a machine, the ego reads it as theft, not addition. People who have their skills, instead of being them, read the same moment as a gift. Same tool, opposite readings. The whole difference sits on the inside.

This is the work coaches have always done. Sitting with someone whose identity is welded to a title, and helping them hold the skill loosely enough to pick up a bigger one.

What survives commoditisation

When the cost of making things falls to almost nothing, value moves to what a prompt cannot produce. Trust built over years. Judgement no dataset holds. The particular taste that makes someone choose your work when a hundred cheaper versions exist.

Nine hundred million people open ChatGPT every week. On the best evidence we have, the thing they reach for most is not code or email. It is advice. Personal, professional, the stuff that keeps people up at night. The largest coaching practice on earth already exists, and there is no coach in it.

For coaches, this is the best news in years. AI can produce coaching shaped text all day. What it cannot produce is presence. The pause before the right question. The years of sitting with someone at their worst and simply staying. Coaching was never the advice. It was what moves between two people when one of them is fully there, and that is the one thing a model cannot reach.

We built CoachNova with coaches, over the past year, on exactly this belief. The coach's judgement is the product. The AI carries the relationship into the space between and beyond sessions. It works under the coach, and never instead of one.

AI is the means, not the goal

Here is the thing most of the noise misses. The opportunity is not AI itself. For the first time, one person with real skill and real taste can take on the kind of ambition that used to need a team of fifty. Some people are pointing that at floating cities and digital twins of the Earth. Honestly, good. The world is more interesting for them.

But the AI is never the achievement. It is the means. The achievement is how much humanity you can carry in with it. The extra decades the island promises, the extra reach this tool promises, they only matter for what you fill them with. AI in service of a person is the whole game. AI in service of itself is just more output on the pile.

So do not build a smaller copy of something that already exists. Build the thing only your own life could have pointed you at. Aim this at people, not away from them.

My three

So I wrote my own list. Three things I want to build before the notice goes up for me. None of them was possible five years ago. Each one needs AI to work. Each one needs a human to matter.

Coaching that stops being a luxury. The best human support has always been rationed. By price, by geography, by luck. A great coach was something most people would never get near. AI is the first thing that can change that maths without deleting the human from it. Not AI instead of the coach. AI under one, so the person who will never afford a standing hour still gets something real. That is the whole of CoachNova. One day I want the same maths to reach therapy.

The reefs. A long stretch on a boat in French Polynesia, then back to Raja Ampat, with whatever I have going into keeping the coral alive. The reefs are dying faster than anyone can watch by hand. That is now as much an AI problem as a marine one. Models that catch bleaching before a diver would. Systems that follow a reef's recovery across thousands of hours of footage no person will ever sit through. I want to be on the boat that carries them.

A shield for girls. Something that protects young girls and young women from the people who exploit them. In too many of the places I love most, that trade runs in plain sight. The hard part was always seeing the pattern in time, across languages and platforms and borders. That is exactly the kind of problem AI is built for, as long as it is pointed at the traffickers and never at the girls. I do not know yet what this one is. I only know it belongs on the list.

Back to the pole

A notice will go up for me one day too. Taped to a pole somewhere, with a number on it. I have stopped caring how big the number is. Medical AI will see to that for all of us anyway. I care about what stands behind it. Three things, if I am lucky and stubborn enough. Built with a tool everyone now holds, pointed where only I would have pointed it.

One test, before you choose your own. If describing it does not make you sound slightly mad, and building it does not scare you a little, you are aiming too low.

The wand is the same in every hand. What you point it at is not.

Louise Doorn is co-founder and CEO of CoachNova.