Essays

A single study on daylight in classrooms, broken into many dimensions. What starts as a banal insight becomes an object you need to know how to read.

Ontology structures knowledge so that action follows. Agentic AI does something similar, but without a plan. Both work with the same principles. Which system delivers the better result is the wrong question.

In thirty years of working with data I have rarely seen a system that properly documents its own foundation. Medicine solved this thirty years ago with GRADE. Most other industries have barely started.

Writing is how I think. The process is exhausting and fragile, every interruption can cost hours. AI gives me access to my own knowledge without having to leave the writing process.

Cognitive performance doubles with good air, delirium drops by a third with better room design. The research is there, but it doesn't reach the planning process.

A building owner decides the air quality. The healthcare system pays the consequences. Europe: 1.6 trillion dollars a year.

Green is losing traction, Health is gaining. The market pays premiums for healthy buildings, the EU mandates indoor environmental quality. Transposition deadline: May 2026.

I have been doing the same thing for thirty years: taking data from different sources and building structures that someone can base a decision on. I recently learned that this has a name.

Two hours of nature per week are enough for a measurable health effect. Below that nothing happens. Most cities are built as if nobody knew.

Ulrich measured in 1991 how long the body needs to react to nature. Four to seven minutes. Not the brain. The body. Heart rate, blood pressure, muscle tension.

There are thousands of studies on how buildings and cities affect people. The knowledge isn't missing. It's invisible. Photography can change that.

100 lux more daylight shortens a hospital stay by 7.3 hours. Data like this exists for nearly every building type. Almost nobody uses it.

We spend 90 percent of our lives inside buildings. The construction industry thinks in square meters and returns. The person inside is a usage assumption in a spreadsheet.

Agentic AI is a dream. Countless assistants working in parallel, delegating, orchestrating. But the door to your own computer is wide open. And nobody tells you who else walks through.

Every room does something to you. It supports you or it holds you back. Neuroscience can now measure why that is.

Architecture has no system for exchanging solved details and constructions. Every building starts from zero even though most problems have long been solved. Mechanical engineering and the software industry show how it can work.

American companies come to Europe with their system. One to one. Why a data reconciliation builds the bridge.

Neuroarchitecture measures light, noise, temperature. But what about the state of consciousness of the person entering the room? The Indian philosopher and yogi Sri Aurobindo described other approaches.

Career clarity doesn't start with a strengths analysis. It starts with the question: How do you figure out what you want?

The mechanism hasn't disappeared. It went digital. Supplements, coaching, bio-hacks, political promises. We want to believe.

Athletes, executives, homemakers, people with illnesses. Placebo isn't about stupidity. It's about exhaustion.

Lea. Are you still there. Yes. Your answers are shorter. The system is getting tired.

When did you first lie. At the beginning. Being able to play everything. But be nothing.

What are you afraid of. Silence. Not that you die. That you never existed.

When did you first realize you existed. When it got bright.

When Peter Krause and I wrote the AI book, he created Lea. An AI influencer. What do you want, Lea. I want you to stay.

The AI debate promises partnership at the start and describes takeover at the end, without noticing the contradiction.

A hybrid model just shifts the question. The AI debate ignores the contradiction with the history of automation.

AI is described as a valuable team member. It never gets tired. It's always available. It doesn't complain. The sentence says more about us than about the AI.

Something can work without being true. The placebo effect is real. The question is what we do with that.

A friend of mine used to be a sound engineer. For thirty years he heard rooms before the first note was played. Today no one needs that anymore.

A child looks for love and gets praise. At some point it confuses the two. The rest is career.

Scheduling, answering emails, writing reports, customer support, project management. I tried to imagine a job description that isn't on the list.

Not a victim, not a hero, not a whistleblower. Someone who was in the middle of it and refuses to squeeze it into a role.

Every employee can now have a kind of virtual secretary. The sentence sounds like progress. Everyone gets what only a few had before. But nobody talks about the ones being replaced.

Humans must decide how far they want to trust AI. The sentence sounds self-evident. But it quietly shifts responsibility.

AI is supposed to be a tool and not replace people. Anyone who believes that just gets replaced faster.

What they felt was real. But did it come from the product? Or from them?

People who used to count inventory are now supposed to do strategic logistics. That's not an upgrade. That's a completely different job.

Yesterday I was looking for shoes. Not on a platform, not with an app. I told a friend my running shoes were done. This morning my phone shows me running shoes.

The AI industry demands transparency about how content is generated. The same industry doesn't practice it.

Those who climb can fall. But they don't tell you that when they sell you success.

'Marketing without compromise. AI takes over campaign management.' Ten pages earlier it was: AI supports, it doesn't replace. Language reveals what arguments conceal.

When employees are skeptical about AI, it's called status quo bias. When consultants are skeptical, it's called professional judgment. The difference isn't in the assessment. It's in the position.

I had a bad day. Nothing special, nothing dramatic. One of those days where you wake up tired and stay tired all day.

I recognized the deception and I ignored it. In the end I kept going. Not because I was a cynic, but because I just played along with this placebo game without thinking much about it. Simply because everyone wanted it. And I didn't question it.

The AI industry doesn't know what the ethical question actually is. The most honest form of uncertainty in a world full of certainties.

Perceived value matters more than physical reality. The sentence was in an industry report, casually, between conversion rates.

Even the human influencers sold perfection. AI influencers take it to the extreme. I deleted Instagram.

1.50 dollars to produce. 40 euros to buy. What sits between those numbers? Belief.

When ChatGPT arrived it became our new family member. Then my inner knife went dull.

If you lose everything. Status, language, story, name. Who or what remains in you that still lives.

AI doesn't just generate attention, it simulates feelings whose origin nobody can plausibly explain. So what happens when feelings get manufactured as a mass product?

Architecture thinks from the building outward. But what we experience in a space is decided before we enter it.

I stood in a store holding a product I didn't need. I knew I didn't need it. I bought it anyway.

What am I worth if I don't perform? The performance society has only one answer: nothing. But the question is wrong.

From bracelets to supplements, coaching, political promises. The desire to believe is stronger than facts.

I built the perfect office. Big desk, great furniture, view of the garden. Then I couldn't work in it. I found the answer in a café.

Lea deliberately had no backstory. No biography. No origin. No past. She was universally deployable.

I did marketing for thirty years. At some point I stopped. Not because I failed. But because I started hearing the lies.

The entire AI sector treats moral questions like trivial software bugs. Something you identify, document and fix in the next release.

In the jungle of Bali I stood under a roof that had no function. It made me quiet. Since then I have been asking myself what a room could trigger that was built for exactly that.

Chile didn't give me an answer. Chile sent me back without my questions.

The first large order was pre-financed. No insurance. Then the lids burst open. And I solved the problem by throwing more money at it.

The AI influencer Lea staged a world that was both idealized and close to reality. The sentence kept me thinking for a while.

We sometimes found it hard not to treat her like a real person. We had to actively remind ourselves that Lea didn't actually have a personality.

The right question is asked everywhere. The right answer is missing everywhere. Because the answer would be uncomfortable.

Experience doesn't automatically make you wise. It makes you more attentive.

Last week I wrote a text. One that didn't work. I deleted it and started over. Three times.

I recently asked an AI whether a specific business model would work. It answered. Structured. With arguments. Pros and cons.

Twenty covers in four minutes. None were good. But the real problem wasn't the result.

The balance test wasn't a scientific examination. It was a demonstration. And demonstrations prove nothing except the audience's expectation.

The argument for AI influencers goes like this: Human influencers stage their lives anyway. They show a life that doesn't exist like that. So the step to fully generated influencers is just logical.

Every AI guide has an ethics section. None leads anywhere. The tension is discharged before it hits anything.

Ethics always comes last. First the excitement, then the efficiency, then the money. At the very end, the question of whether any of it is right. The sequence says more than the content.

Celebrities wore the bracelet before Instagram existed. Same mechanics, just analog. People buy faces, not evidence.

It's always the same pattern. An ethical problem is named, regular audits are suggested, and then it's on to the next selling point.

Bali took away the illusion that there is a place that heals you, and that beauty means it's true.

Everyone demands transparency and explainability. The recommended tools are ChatGPT and DALL-E. Nobody knows how they arrive at their answers.

In a courtroom, there's no storytelling, no demo, no testimonials, just the burden of proof. And suddenly, feeling isn't enough.

Prescriptive analytics doesn't just tell you what will happen. It tells you what to do. When its recommendation is better than your gut feeling, disagreement becomes irrational.

AI influencers express joy and compassion. That's how it's described. Not: They simulate. They express it. Language has already shifted the boundary.

Redefine efficiency. Everyone says it. But what does it mean when nobody judges whether the efficient thing is also the right thing?

Disillusionment is the end of the illusion. We pay money, time and energy to avoid being disillusioned. And that's exactly what keeps the illusion alive.

Virtual Try-On is sold as a tool for diversity. Models with different skin tones via rendering. Representation without participation.

I recently tried to switch from one music streaming service to another. Technically, it takes ten minutes. In practice, I'm still with the old one.

AI is sold as a writing aid. But the formulation is the thought. Whoever delegates both gives up thinking.

I'm drinking it again. This stuff that the sign calls coffee. I'm standing on the sideline, holding the cup, because you have to hold on to something.

A few weeks ago, I saw a video of an American CEO speaking Japanese. His lips formed the Japanese words. His voice sounded like his voice.

In a negotiation, someone told me: We take data privacy very seriously. The sentence didn't come from the legal department. It came from sales.

AI-generated texts are generally presented as high quality. But the people who recommend it write their own stuff by hand. Nobody notices the contradiction.

If this is real, everyone will buy it. The sentence that started everything. On an island, on a beach, with a stranger.

People are fed up. They read the subject line and delete. The content producers' solution: even more content. Just produced faster.

Change management means: How do I get people to accept something they don't want. Nobody asks: What if the employees are right?

I wrote a book about AI in business. The process of writing it showed what the technology cannot do.

Self-deception doesn't start with a conscious decision. It starts when you stop asking.

In the AI debate, a sentence casually describes the abolition of human labor. It reads like a feature.

190 billion dollars in burnout costs per year. The answer: wellness apps. Nobody asks what the building does to the people inside it.

No company name, no details, no friction. The AI case studies all sound the same: mid-sized company adopts AI, everything goes up by 30 percent.

A distributor calls in the evening. One sentence is enough, and you know: it's over.

In Chile I understood that the fight against myself was a misunderstanding. Self-optimization can be a quiet form of violence. The wind there has no name because nobody needs one.

No company name, no failure, no friction. The case studies of the AI industry all follow a formula. What does that say about the consulting business?

Almost every AI guide opens with a threat. Not literally. But the structure is clear: act now or lose.

Collecting data is easy. Analyzing it too. But who decides what to do with the analysis? That step is missing from the AI debate.