Spatial Web operates five companies that turn real estate into machine-readable space — and researches the models that will read it.
Five companies sell spatial products into real estate — in production, with paying customers. The business funds everything downstream.
Every job adds to one corpus: photoreal, measured, labeled interiors of real, lived-in homes — joined to verified market records.
The corpus feeds work on spatial LLMs — models that read rooms the way language models read text.
Language models learned from a web of text. The next models need a web of space — and someone has to build it.
What listing capture actually contains — imagery, geometry, semantics, and market ground truth — and why lived-in homes are the missing data in physical AI.
A position paper on the architectures and training data required for language models grounded in the built environment.
Market records as free ground truth: using ownership, valuation, and listing metadata to supervise indoor scene understanding.
Papers publish here first. Labs and robotics teams can request early drafts and corpus documentation at inquiries@spatialweb.io.