I photographed the future standing behind the past

Across a Copenhagen harbour, a nineteenth century spire stands in front of a power plant you can ski down. That same gap is now opening between the AI we use today and the intelligence coming next, and it is why you build for it before it arrives.

Yesterday I stood across the harbour in Copenhagen and took this photograph. Look at it for a moment before you read on.

In front, a 19th century building. Red tiled roofs, and at its end a dark bell-shaped spire gone almost black with age, the kind of craft that took a generation of hands to raise. Behind it, jarring against all of that, a machine of brushed steel and glass shears up into the sky like a mountain that should not be there. One was built to last a hundred years. The other looks like it landed last night. They have no business sharing a skyline, and yet somehow they belong together.

The machine is CopenHill. Beneath that slope it burns much of the city's waste each year and turns it into clean power and heat for hundreds of thousands of homes. On its roof there is a ski run, hiking trails, and one of the tallest climbing walls in the world down one flank. Bjarke Ingels, who won the competition to build it in 2011, calls the idea hedonistic sustainability. A sustainable city, he says, does more than serve the planet. It is more enjoyable for the people who live in it.

Here is the detail I cannot let go of. At the time, the whole idea was thought impossible. A power plant is something you hide behind a fence. Nobody climbs one for joy. They built it anyway. It opened in 2019, and years later it still stands almost alone, rarely equalled anywhere. They had nothing to copy and no proof it would hold.

A power plant people would climb for joy was thought impossible. They built it, and years on it still stands almost alone.

That is what being first actually costs. You move before the map exists, because you are the one drawing it.

We are all standing in front of that same skyline now. In the foreground, the world we know, the craft and judgement of a career. Rising behind it, a machine most of us do not yet fully understand. Our first instinct, when the new thing appears, is to treat it as a threat to the old and to choose between them.

The Danes refuse to choose. One of the happiest countries on earth, and one of the most productive. A state oil company there became the world leader in offshore wind, because it declined to accept that the two could not be one.

I have spent 25 years in hospitality, an industry built on those same false choices. Cost or experience. Efficiency or warmth. A few weeks ago I wrote about four teams in one hotel who were each completely right and still lost money, because no one turned their four right answers into a single decision. That is how it usually goes. A choice looks correct inside the team that makes it, and quietly creates a problem two corridors away, and nobody sees the whole shape of it until the numbers come in at the end of the month.

Now look again at the machine rising behind us. The large language models we use today are remarkable, and still only the foreground. An LLM answers the question you asked. What comes next, what the field calls AGI, or Artificial General Intelligence, is different in kind. Think of the difference between a tool that does exactly what you tell it and a seasoned operator who understands the whole business, notices what you missed, and acts on their own judgement. It would hold the entire hotel in mind at once, connect two facts nobody put in the same room, and act while the night can still be saved, long before it would ever surface in a quarterly review.

An LLM answers the question you asked. AGI will answer the question you did not know to ask.

The experts cannot agree on when it lands. I have stopped finding the date interesting. You begin long before the machine exists. You build the institution around the belief that it will arrive, so that on the day it does you are already shaped to use it.

That is what we are building with PHAL, and it goes beyond anything most hospitality chains run today. The industry has always measured four separate truths. Performance, Profit, Preference, People. Four systems, read by four teams, and no group has ever held all four in a single view with decision making capability, because until now the technology could not carry them together. That is what we built. Where the industry reaches for another dashboard, we built one layer that reads all four Ps at once, so four right answers finally become one right decision. Today it runs on agentic AI. We are building the foundation now, so that as the intelligence grows into what the field calls AGI, the operator is already shaped to use it. We started before it was proven, because someone has to move first.

The old building and the new machine share one skyline, and neither had to fall for the other to rise. The impossible is rarely a wall. It is almost always just a combination nobody has yet dared to make.

I am grateful for the keen interest this has already drawn, and I look forward to building it alongside the groups with the courage to move first.