Physical AI for the Built Environment #SFTechWeek

Thursday, Oct 8

5:30pm

PTET

Hosted by

ᵔ◡ᵔ

◉‿◉

Stand hosts an evening at its San Francisco headquarters on Physics-Informed AI for the built environment. The evening covers world models that simulate how fire and wind physically destroy structures, and agents that act on those simulations to recommend specific fixes, like clearing brush or retrofitting a wall, and price insurance accordingly. Doors open to a happy hour with live simulations running for guests to explore. Watch a wildfire move across a real home, wind fields wrap a structure, a property before and after hardening. These are the same simulations Stand ran against hundreds of homes in recent California wildfires, predicting which ones survived. Stand engineers will walk guests through the production pipeline behind each demo and answer questions. The program then turns to the two halves of Stand's thesis: the business it built, and the deep tech underneath it. Dan Preston (Co-founder & CEO, Stand) delivers a keynote on Stand’s vision: merging capital markets, real-world protection, and physics simulation into a single business, and how that combination has powered its rapid growth. Matt DiStefano (Chief Science Officer & founding team member, Stand) follows with an introduction to the technology: how the Stand World Model learns physics, how it was validated against real catastrophes, and what it took to put it into production. Matt then leads a panel on what stands between Physics-Informed AI and wide deployment, and which capabilities are production-ready today. Joining him are Dr. James O'Brien (Professor, UC Berkeley; expert in physics simulation and AI; Academy Award winner for the graphics used in 200+ films and games; co-founder of KronosAI and Get Klothed) Dr. Ahmad Peyvan (Assistant Professor, Vanderbilt University; previously Brown; building scientific machine learning for high-speed and hypersonic physics) Daniel Martinez-Gonzalez (Machine Learning Engineer at Stand; previously NASA Ames and the U.S. Army, where he developed neural-network digital twins for rotorcraft). Discussion topics will include: Hybrid architectures: where does fast AI approximation end and full physics simulation begin, and does that line move over time? The production gap: which AI surrogates work in a demo but break on real-world cases, and why? Generalization limits: the best methods available today for handling instability, and for generalizing across new geometries and boundary conditions. The program closes with a live demonstration of Stand's agents interacting with its World Model, optimizing the mitigation and discount for specific homes. This is the technology now underwriting over $8 billion in insured home value, and what lets Stand protect and insure what others can't. Networking with the panelists, engineers, and builders continues afterward. This event is a part of #SFTechWeek—a week of events hosted by VCs and startups to bring together the tech ecosystem. Learn more at www.tech-week.com.

Guest List

39 on the list

View all

⚆◟⚆

•ᴥ•

◉‿◉