Andy Elder and Arjun Mathur, who were roommates at Stanford and both studied computer science, spent roughly a decade working at Palantir. At Palantir, Elder led product design for an 800-person team building software for U.S. defense and intelligence customers, while Mathur served as a senior vice president in the company’s commercial practice. Early last year they decided to start a company together.
Why they targeted the built environment
Neither founder has a direct background in architecture, but both are concerned about a long-term trend: fewer buildings are being constructed in the United States today than a few decades ago, while development costs have risen. Mathur told TechCrunch, “We’ve gone from dial-up internet to superintelligent AI, but our ability to build the physical world around us has actually decreased.”
What Dream does and how it works
Their startup, Dream, is developing built-world models—AI systems intended to automate building design and engineering across the architecture, engineering, and construction (AEC) sector. The company says its aim is to replace slow, legacy AEC tools such as Autodesk’s AutoCAD and Revit.
Unlike a standard large language model that can produce an image of a building but not ensure structural soundness, Dream’s approach is grounded in physical rules and construction standards. The company trained its world model on more than 100 million square feet of building data gathered from permit archives across the United States.
Elder said, “Our model actually understands what’s within those walls: what materials are used, what the beams are made of, and how much weight they can support.” Dream calls its method “structural cognition,” meaning the model generates structurally precise designs that account for internal building systems like plumbing and HVAC.
Early customers and deployments
According to the founders, Dream has already begun selling its product to AEC professionals and developers who use the model for conceptual design and cost estimation. The platform is deployed across 10 active projects in Texas and California.
Other startups are also attempting to modernize AEC software using AI—examples include Motif, Snaptrude, and Arcol—but Elder and Mathur contend they are the only company that has developed a world model capable of designing whole buildings instantly.
Funding and investor perspective
Dream emerged from stealth with $17 million in seed funding led by Andreessen Horowitz, with participation from SV Angel, XYZ, and Zigg Capital. Kimberly Tan, a partner at Andreessen Horowitz, said the firm invested in Dream more than a year ago when the startup was still an idea, and praised the founders’ technical approach: “I have not seen anybody take the technical approach. They’re doing everything based on their own models.”
Why it matters
If Dream’s physics- and standards-based world models work at scale, they could speed up building design, reduce errors, and lower development costs in an industry that has lagged behind other sectors in productivity gains. The company’s initial deployments and investor support are early indicators, but broader industry adoption remains to be seen.
Summary
Andy Elder and Arjun Mathur, former Palantir leaders, founded Dream to build AI-driven world models for structural and architectural design. The startup has raised $17 million in seed funding, trained its models on over 100 million square feet of permit data, and is working on 10 active projects in Texas and California.
