AI Construction Engineering Startup Spacial Taps NYU Prof Ravid Shwartz-Ziv

Spacial, an artificial intelligence-powered engineering platform for residential construction, announced that Ravid Shwartz-Ziv has joined the company as a scientific adviser.

An assistant professor and faculty fellow at New York University’s Center for Data Science, Shwartz-Ziv leads large language model (LLM) research focused on using information theory to explain what neural networks actually learn. He will work closely with Spacial’s research and engineering teams, advise on major technology decisions for the platform, and assist the founders in establishing its research road map.

Spacial’s platform converts two-dimensional architectural drawings into 3D, code-compliant construction documents with built-in design conflict detection, architectural detailing and local code validation. Founded in 2024, the Palo Alto, Calif.-based startup can turn 2D plans into stamped structural and mechanical, electrical and plumbing (MEP) permit sets in days using AI, plus licensed engineer review.

“Before an engineer will stamp what our model produces, we need real confidence in

what it’s learned, not just how well it performs,” Maor Greenberg, co-founder and CEO of Spacial, said in announcing Shwartz-Ziv’s hire. “The construction industry doesn’t have a

shared standard for proving that an AI-generated plan is accurate and code-compliant. Closing that gap is exactly why we brought Ravid on. He brings deep expertise in evaluating AI systems, and that’s exactly the rigor Spacial needs to get structural AI right.”

Shwartz-Ziv has spent several years benchmarking and evaluating AI systems, which is relevant to a problem the construction industry hasn’t solved: There is no shared standard for measuring whether an AI-generated plan is accurate, code-compliant and internallycoordinated.

In addition, Shwartz-Ziv will guide the evolution of Spacial’s core technology, a model trained to understand buildings as systems of physical relationships. The goal of which is to produce permit-ready engineering documents for construction projects, an issue that is closely aligned with Shwartz-Ziv’s research. He earned a doctorate in brain science and computer science from Hebrew University of Jerusalem and has worked as an AI researcher at Meta, according to his LinkedIn profile.

“Permitting is a hard, practical problem, and Spacial is taking a serious technical

approach to it,” Shwartz-Ziv said in a statement. “Confirming that a model has genuinely learned how buildings work as physical systems, not just that it can produce a plan that looks right, is exactly the kind of question my work on information theory and model evaluation was built to answer. I look forward to supporting the team as an adviser.”

In 2025, Spacial secured $10 million in funding by demonstrating its ability to document progress, reduce rework, and compress pay-app cycles, according to the Center for Real Estate and Technology & Innovation (CRETI). The CRETI report found that the underlying pattern is accelerating: Investors want construction intelligence that produces lender-acceptable evidence, not optional workflow enhancements.

Philip Russo can be reached at prusso@commercialobserver.com.