t>

This startup just got funded to take another go at the same problem: affordable housing design


Nick Donahue’s parents were in the construction industry, which means he spent his childhood hearing about the US construction industry. His father built homes for major builders, and his mother sold to big box builders on the East Coast.

Donahue was very interested in why designing a home was so expensive and time-consuming, and why so many people had to settle for what the manufacturers offered that year. So after dropping out of NC State and moving to the Bay Area, he ended up doing what you might expect a college dropout to do in San Francisco: He started a company to fix it.

That effort, Atmos, went through Y Combinator, he earned $20 million from investors such as Khosla Ventures and Sam Altman, and tried to use technology to change the construction process. It had staff developers working with customers while software was running in the background. It grew to 40 people and $7 million in capital, and built $200 million worth of homes and built 50.

All of which sounds great until you hear Donahue explain it. “It became a very active business,” he told me on a Zoom phone call last week. “It’s like a beautiful construction company.”

They didn’t replace people, in other words. Then the Federal Reserve started raising interest rates, and suddenly customers who had spent months building their dream homes couldn’t afford them. Nine months ago, Donahue closed.

This is where most startups can take a break, maybe write a few LinkedIn posts about what they’ve learned. Instead, Donahue rejoined and started another company.

Written now they are almost five months old, and all Atmos was not. There are no manufacturers on staff. There are no operational problems. The only AI-powered software that creates floor plans with exterior designs in minutes. You tell him what you want – bedrooms, square footage, whatever – and he spits out five designs. Don’t like them? You can make five more and keep going until something clicks.

Techcrunch event

San Francisco
| |
October 13-15, 2026

Currently, Drafted has six employees, four from Atmos, and raised $1.65 million for a $35 million after investment from Bill Clerico, Stripe’s Patrick Collison, Jack Altman, Josh Buckley, and Warriors player Moses Moody.

Clerico led the group because he also had angels at Atmos and watched Donahue’s house survive despite rising interest rates. When Donahue told him about the new coffee company, Clerico didn’t need to be convinced. “Nick, please take our money,” he apparently said over and over for two weeks until Donahue agreed.

The line is straight. Right now, if you want a custom home, you have two options: hire an architect (expensive, slow), or buy a template plan online (cheap, unpredictable). The graphics are central, providing customization for the template prices. A complete system costs between $1,000 and $2,000.

Finance works because Drafted built its own AI model, trained on real home plans from homes that have been built and let. Performance constraints are taken into account, and Donahue says the special model doesn’t cost anything: two-tenths of a penny per floor plan, compared to 13 cents for a standard AI.

Plans are limited to single-story homes right now, but more and more are coming. The big question is whether there is a market for this.

The numbers are not great. Of the million new homes built in America each year, only 300,000 are custom built. Most people buy existing homes or choose whatever homes the major builders are offering.

Clerico’s argument is that this is a chicken and egg problem. Make your design cheap and fast enough, and more people will do it. Donahue compares it to Uber, which not only replaced taxis but created a ride-hailing service for many people. Clerico said: “There is no reason in the future that everyone should not have a custom home.

Or maybe more Americans will be consumers who take what’s available. The housing market has a long history to restore confusion.

There is also the question of “moat”. When asked what should stop an LLM player or another independent player from buying the same datasets and creating the same thing, Donahue talks about brand, pointing to his colleague David Holz, who started the video and graphic design of AI clothing, Midjourney. Although there are many new models being introduced, the usage of Midjourney is not moving, Holz told Donahue; his clients keep coming back to create AI images.

Likewise, Donahue thinks if they move fast enough and attract enough customers, Listings can become a place where people can make a home.

Time will tell. Since opening to the public, the outfit has seen nearly 1,000 daily users. Not huge numbers, but they show the steady growth of such a small thing.

Meanwhile, Donahue has something valuable that can give the Journal an edge: a deep understanding of the problem and the experience gained from doing it before.

Pictured above, the Drafted team, left to right: Martynas Pocius, Albert Chiu, Martina Cheru, Carson Poole, Stephen Chou, and Nick Donahue.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *