I co-founded Opendoor ($10B+ rev). Today, I’m launching Summation. We raised $35M from Benchmark & Kleiner Perkins to build an AI analyst you can trust. Here’s where Claude and ChatGPT fall short: 🧵

I spent 24 hours inside America's newest car-free neighborhood, with the man who designed it. @ryanmjohnson It made me realize how broken everywhere else actually is – and how American cities are about to change forever. What I learned is that autonomous cars will reduce the need for the majority of parking. When you remove parking lots and garages from a neighborhood, you free up an enormous amount of land. That land can become courtyards, gardens, restaurants, and shops all within walking distance of every front door. The company that made this town, Culdesac built an entire neighborhood in Tempe, Arizona on this idea. And the result is something that could be the blueprint for how we design our neighborhoods once AVs become fully adopted. The insight is simple: parking doesn't just take up space, it kills the energy of a place. Remove it and you get density, walkability, and community back. Now they're scaling it. More cities, more neighborhoods, a different vision for how Americans live. Most people think walkable cities are a European thing — too old, too dense, impossible to replicate here. Culdesac is proving that's wrong, one neighborhood at a time and giving us insight into how we can develop neighborhoods for AV age. @culdesac

What many people discussing this fail to realise is that the Netherlands is one of the best places in the world to cycle and to drive a car.

Today, I’m launching my newco, NavigateAI. We are short hundreds of thousands of skilled workers and we're on a mission to give every field worker an AI copilot, so they can build faster and better when we need it most. navigate.ai/blog/2026-05-26-…

Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.

Electric bike and scooter rental startup Lime filed for an IPO, showing growing revenue and an expanding user base bloomberg.com/news/articles/…

Nike's killing it with this ad in SoHo New York City

The building will have no amenity space. The neighborhood itself is the amenity.

Opendoor is getting attention for offering mortgage rates that look "below market" and I want to talk about it. This isn't some magic trick. It's actually pretty basic. Here's how we do it: Opendoor mortgage rates aren't marked up. The end. See, when people talk about "market rates" for mortgages, they telling you about the rates they see online from their lender, or from Mortgage News Daily, or some other source. Remember: those rates include 350 basis points of markup on average, based on self-reported data to the Mortgage Bankers Association. 350 basis points is not nothing. As a rough rule of thumb, every 100 basis points markup raises a consumer's mortgage rate by 0.25 percentage points. So, let's all acknowledge that "market rates" in mortgage reflect 350 basis points of markup, which raises a customer's mortgage rate by roughly 0.875. Opendoor changed that. Our mortgage rates are what happens when you take that markup out. It's like what E*TRADE did for stocks. In the 1980s, the market price of a stock was whatever its price was plus whatever your broker charged. It's why every broker had a different price. Today, the price of a stock is the same everywhere. So if Opendoor's mortgage rates look "below market" to you, they're actually not. This is just the first time you're seeing mortgage rates without a massive markup. More here: opendoor.com/articles/why-mo…

Some reflections after returning from a 15 year twitter sabbatical (#82 is a real doozy)