Image-based registrations are live in proprty.ai

Doing a registration in proprty.ai used to mean walking the property and typing in what you saw. Now you can do it with a camera. A new feature in the app lets users take a photo of a building part, and the AI predicts the material, installation year, condition and remaining lifetime.

Customers build maintenance plans in proprty.ai in three ways. The first is to rely on the AI's estimates from public building data, which are already accurate enough that many use them directly in budget planning. The second is to do fully manual registrations on every building part for maximum precision. The third, and most common, is to mix the two.

"Most customers land somewhere in the middle," says Mikkel Jensen, co-founder and CDSO at proprty.ai. "They trust the AI estimates for the bulk of the portfolio, and they do registrations where it matters most. Image-based registrations make that middle path a lot easier to take."

What changes for the user

The more registrations you make, the more your plan reflects the actual state of each building part. Image-based registrations make that work faster, and lower the bar for who can do it.

Until now, a registration meant entering material, installation year, condition and remaining lifetime by hand on every building part. Compared to a traditional condition assessment, that was already faster, more streamlined and produced structured data from day one. The image flow takes the manual entry out on top of that. The user takes a photo, the AI proposes a full registration, the user reviews and adjusts.

"If you have an image of a building part, you have a lot more information than we can ever get from public building data," says Mikkel Jensen. "That gives you a much better starting point, and in most cases the prediction is already close enough that you only need to verify it."

Built on what proprty.ai already knows

The image prediction is not a generic guess. It plugs into the same structure that powers the rest of proprty.ai: a material list per building part, average lifespan values per material and the mapping from condition scores to remaining lifetimes.

"The prediction picks from our own material list, which feeds into the average lifespan for that building part and material, which then anchors the condition score and the remaining lifetime," says Mikkel Jensen. "Everything we have built over the past years is what makes this work."

More improvements to come

The image flow is one of several changes on the way to a registration experience that requires less typing and less expertise. The user still reviews every prediction, and that part is not going away.

"It takes less skill to get a registration right when you start from an image," says Mikkel Jensen. "And it frees up the user's attention for the parts of the building that actually need a second look."

proprty.ai uses domain-specific AI to predict building condition, remaining lifetimes and cost across large portfolios. The result is a continuously updated decision foundation for maintenance and investment, with condition, cost and CO₂ connected in one place. Customers across Denmark, Norway, Switzerland and Germany manage more than 40 million m² in proprty.ai.