Post by Mattias Petter Johansson

Paw prints in the sand in Erg Chigaga, Morocco. Susanne Drakborg photographs them because she's curious, and uploads the photo. The AI says: Fennec Fox. How does it know this? Someone, somewhere, looked at an image of fox prints and typed the words 'Fennec Fox', along with their reasoning. Then someone else did it. And again. Thousands of times, until a model could recognize the shape. The model is only as good as what someone bothered to write down. Next week I'm starting as Solutions Architect at Hubexo. If I was me three months ago, I'd be wondering—why, in the middle of the raging AI storm, is MPJ joining some weirdly named data company in construction? The short answer: my friend Zoey Tsopela prompted me one dark autumn evening: "Construction, man. Concrete. Steel." Credit also to Oliver Finnman and Oliver Aulin at Detechtiv, without whom Hubexo would've flown under my radar. The deeper answer starts in 1936, when Bengt Wigart established *Byggnadsupplysning* in Stockholm. His method was purely manual: cycling to municipal offices to collect building permits, then selling them to contractors. The value was not in the public data itself, but in the labor of aggregating it. Today, ninety years later, 2,500 researchers in Ljusdal, Manila, Vietnam, the UK make over a million verification calls a year. They phone developers, architects, contractors. They fact-check what the algorithms extract. They write it down. It's 2015, a dark evening near Christmas. I'm sitting at a table in a mezcal bar (whose name would downrank this post, but it's around the corner from the the Spotify Gothenburg office) Fredrik Norén mused: "us developers are essentially just some weird translators." That sentence stuck to my brain like flypaper, nagging me for years. The work getting easier, not because I improved—because it demanded less. I needed to spend less time in computation and more in what was being computed. Oscar Carlsson taught me data. In data, I was effectively junior again—much more than I imagined. Kevin Goldsmith mentored me in leadership, which was newer still. Both made the shift possible. Five uncomfortable years later, I'm joining a company that's been annotating the built world since before computers existed. Twenty buckets of data scattered across ninety years. Second week of the job, I'm flying to Jakarta—and somewhere between time zones it occurs to me that this is what the work has always been. Someone cycling to city hall. Someone picking up a phone. Someone landing in a place where the data can be found. The paw prints are still in the sand.