Every Building Needs a World Model

Your Roomba maps your apartment before it mops it. That little robot knows it needs to understand the space before it can do anything useful in it. Map first, then act.

Now think about the building you're in right now. How much does any machine actually understand about it? Almost nothing. The most sophisticated tool managing most buildings is still a spreadsheet.

Buildings are our outer shell

Buildings are extensions of ourselves. They're the shell between us and the outside world. We spend about 90% of our lives inside them. They regulate our temperature, protect our stuff, shape our days in ways we barely notice until something breaks.

And yet they're invisible to the machines we're building. Self-driving cars have centimeter-accurate maps of every road. Your building doesn't have a map a robot could read.

Simple tools, complex understanding

You can already capture a building with tools everyone has. A phone with LiDAR. A video walkthrough. A few photos from different angles. The capture part is solved.

What's not solved is turning that raw scan into something meaningful. A room isn't just walls and a ceiling. It's a space with history: who uses it, how they've made it theirs, the paint they chose, the lighting they prefer, the layout that reflects how they actually live or work. A building is structure, but it's also the preferences and habits of the people inside it.

Understanding a space means knowing all of that. Not just what's there, but why it's there and how it's used.

That's the gap between a 3D scan and a world model. One gives you shape, the other gives you meaning.

What we're building

Pascal creates world models for buildings. We take the physical space and turn it into a structured model that machines can work with.

It starts with geometry. A phone with LiDAR, a video walkthrough, a set of photos. Simple capture tools that exist today, pointed at the building as it actually is, not as it was drawn on a blueprint years ago.

Then we layer in meaning and live data. Power consumption from smart meters. Images from cameras already installed. IoT sensors tracking the basics. Objects get identified, relationships get mapped. The HVAC unit on the roof is connected to the panel that powers it, the zones it serves, the history that tells you how it's been behaving.

The result is something nice to look at for a human, but more importantly, something complete enough for a machine to reason about.

Why this matters now

Machines are coming indoors.

AI agents are already managing building systems, catching problems early, coordinating repairs. That's what Pascal does today. The next wave is physical. Delivery robots, inspection drones, maintenance machines. They're already in warehouses and factories. Buildings are next.

When they arrive, they'll need what self-driving cars already have: a structured understanding of the space they're operating in. Which door leads where. What's behind that wall. Why this room runs warm. You can't get that from a floor plan.

You get it from a world model.

Every building deserves a map

We've mapped Mars. We've mapped the ocean floor. We've mapped every road on the planet so well that your phone can navigate a foreign city in real time.

The buildings we live and work in every day? Most of them have never been properly mapped at all.

The building is your outer shell. It deserves to know what it's made of.

---

Julien Brissonneau, CEO — Aymeric Rabot, CTO pascal.app