Yes, we're going to solve the housing shortage with AI!

Artificial Intelligence is a tool that makes all your work easier, or so people think. But it's not that easy, thinks Pim van Meer. "Without a clear data set and testable goals, AI will mainly speed up chaos," he argues. And he shows you how you can make AI work for you.

I am a change manager. And yet I can't stand change. For years, AI and voice tools were like toys next to my keyboard: fun to have around, not something to lean on. Until I literally went down, broke my collarbone and stubbornly kept trying to type for two days. I have been able to type blind since elementary school - fast, too - but I talk five times as fast. It wasn't until I couldn't anymore that I really had to switch gears.

Since then, my pieces are not created because I am romantically talking to a phone and it is perfect in one go. A speech tool only puts down what I say. A language model doesn't figure out what I want to say. The gain is in combination and rhythm: clever stacking, not worshiping one miracle tool.

This is how I work now - and why it works

I start with rough clay: in one take I speak the core, an anecdote and the filleting moment. No prettiness, but speed. Then I ask AI for order and headwinds: rank angles. Not to replace me, but to speed me up and push me back where it's thin. Then I go back to voice: sentences that read stiffly, I re-voice. Speaking aloud arranges, points out gaps, alternative angles force simplicity; you hear immediately where you falter. Where I can, I hook data to my claims: lead times, revision loops, data completeness. And then the kill moment: one hard delete. Rather short and sharp than long and lukewarm. Result: a column every week, with no evening marathons. Not because tools are "smart," but because my work order is smarter.

What does this say about change? That we are all sheep who like to walk with the herd - me included. Familiar feels safe, even with someone who change has on his business card. The biggest lesson of my collarbone: real change often comes not from courage, but from limitation. Only when something is no longer possible, does it create room for something better. Call it constraint-driven innovation.

It works the same way on projects. As long as "we can still do it the old way", habit wins. The tipping point is in rhythm with choices: once every two or three weeks in the design team or construction team meeting looking at the exact same meters and logging decisions with stop, go or fix. Not because everyone is suddenly rabidly enthusiastic, but because repeatable choices lower the threshold. Discipline > decibels.

What AI does and does not do for you

AI is not an autopilot. It is a copilot that can do three things very well - if you do your homework. Reading and explaining: with one source of truth (model + data fields), a digital thinking service can answer questions and show everyone the same facts. Testing: with threshold values per room (daylight, sound, WWS, accessibility), a quality agent can preview before the meeting. Summarize: with a tight decision log, a decision-buddy summarizes the building team meeting, including owner and follow-up. Sounds big, starts small: one use case, one dataset, one rhythm. Not a thousand variants; two that can get through the gate.

The pitfalls I fall into

Pace: AI tempts to sprint beyond the adoption capacity of the team; flag-bearers then burn out. Perfection: sealing everything up first is an elegant way not to start. No harvest: rushing through without a mini-ritual. I'm bad at that. Four minutes of harvest at the end of a session does wonders for motivation.

What can you do today without a task force or tender party? Start painfully small. One building type, five data fields, one threshold per field, one rhythm. Let AI help only where the noise is: reading, testing, summarizing. Everything beyond that remains just craftsmanship. And if it chafes in the meeting, remember: dosage wins over dramatics. You won't convince anyone by talking louder than their adoptability allows - even if your model is perfect.

Small confession in conclusion

I once said in a session: Give me five extra dashboards and I will arrange support. Nonsense. Support does not follow graphs; support follows behavior. The behavior you practice together every two or three weeks. That's where AI fits in: as a silent engine behind clearer choices, not as a siren on the roof.

The filleting moment

AI will automatically speed things up.
No. Rhythm provides acceleration. AI augments what you organize: without a clear data set, information needs, verifiable goals and decision cadence, you mainly get faster chaos. With minimal information needs and fixed stop-go-fix, change becomes boring - and inevitable for that very reason. AI helps you in your work if you know how to translate work digitally.

About Pim's digital maze

In this column, Pim takes you into the sometimes wonderful, tangled but rapidly changing world of digitization. He draws on his experiences as director of digitalization at VORM. Pim is outspoken, critical, but above all wants to help you. Are you stuck in the digital maze? Pim helps you find the way out.... Do you have a question for Pim or are you looking to get in touch with him? Follow Pim via LinkedIN.