AI makes starting a company easier than getting a job, but solo founders have no fallback
Durable hit a million users in three months, but its founder says he took a detour: website generation was not the endgame — handing the entire back office of the service industry to AI is. Being a solo founder isn't more freedom; it's carrying every decision alone.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
3.5 million sign-ups, and over half are brand-new businesses
About 3.5 million people have signed up on Durable to start a business, and more than half of them are net new — not existing businesses coming online, but people starting a company for the first time. These people aren't in the brainstorming phase; they already have an idea and possibly some revenue, and the first thing they do on Durable is build an online presence, do SEO and GEO, get found on ChatGPT and Google, and start running ads and following up with customers. Clift says this showed him a mass of businesses he "could never have dreamed existed," like decorating people's porches for Halloween, which can do several hundred thousand dollars a year.
— James CliftThe AI website generator was a detour, not the endgame
Durable's original vision was "Shopify for the service industry," close to a franchise model: the franchisee collects $100,000, Durable takes $10,000, and provides the same leads, website and marketing system. After GPT-2.5 appeared, they ran a one-week internal hack week, built an AI website generator, launched it in 2022, and it went viral, going from zero to about a million users in three months and bringing in meaningful revenue. But they spent a year and a half on that product, and Clift now calls it an "interesting distraction" — a good business, but not the market they wanted to be in. Four years later they returned to the original vision: an AI version of an "out-of-the-box business" for service business owners.
— James CliftUsers don't want buttons, they want outcomes
Clift deliberately avoids the UI details in customer feedback: everyone says they want to change this button, add this component, customize that, but what they actually want is a better-looking, better-performing website. Likewise, appointment scheduling can fall down the rabbit hole of time zones and multiple calendars, but what people actually want is "jobs getting booked." His conclusion: building UI is now extremely low-value, first because AI makes UI easy to build, and second because nobody wants to click through a pile of things anymore — if users never log into Durable and just get one text message a day, that's a better experience. The product should be input and output: describe the problem you want solved, and it gets solved.
— James CliftYou should now deliberately pick harder problems
Clift says that since models will keep getting stronger and will make your product obsolete ("it may already be obsolete, and it will be obsolete again at least once within six months"), you should deliberately choose harder problems. Durable has millions of people actively starting new businesses, and making them succeed is an extremely hard problem, but once solved it corresponds to trillions of dollars in GMV and economic value. He also mentions team size: still only six engineers, but "six engineers are roughly equivalent to sixty agents," so adding features is no longer scary, and the real risk becomes "are you building the right thing for the right customer."
— James CliftSolo founding is about shedding responsibility to other people
Clift says he didn't deliberately choose to found solo; rather, his first two partnerships with co-founders both got stuck on the same thing: mismatched runway and risk. He couldn't help calculating the other person's opportunity cost — "how much they could make at Amazon, how much they could make consulting" — and by rational calculation you should never start a software company. More critical is the psychological burden: with even one co-founder, he felt he had to keep that person happy and take care of them, effectively carrying that person's family's livelihood on his shoulders. Founding solo removed that pressure — "now it's all on me alone" — and he can think only about the business, do creative work more freely, and if he fails, only he is affected.
— James CliftBuilt in Vancouver, funded in San Francisco
Clift's judgment: to win a venture-backed, outsized outcome, you probably still have to be in San Francisco; but Vancouver is a great place to build, and San Francisco is a great place to raise money and get inspired. Every time he goes to San Francisco he thinks "we should work harder, faster, be more ambitious," but that energy also often gets in the way of actually building product and understanding customers, producing that lost founder who is "caffeinated, trying a million things, gripping the handlebars and having no idea what they're doing." His approach: the company DNA, ambition and fundraising are Silicon Valley-style, but the customers aren't in Silicon Valley — they're in the middle of America, internationally, and they're solo owners.
— James CliftDoor-to-door knocking teaches you rejection
In college Clift cleaned windows and gutters, selling door to door. He learned two things: first, desensitization to rejection — nobody yells at you for knocking and offering help, and they forget you two days or even two minutes later; second, volume. Knock on enough doors and you'll close, and then a neighborhood "flips": one house does it, puts up a sign, and the neighbors all follow. He directly analogizes this to fundraising and cold outreach emails: the worst outcome is that they don't read it, and that's probably what will happen. He also says that if building becomes easier because of AI, then sales becomes more valuable.
— James CliftEarning your first dollar yourself is a worldview unlock
Clift thinks one of the most interesting parts of the human experience is "earning money yourself," and not even in a financial sense. The world is designed around employment, and profit flows to business owners; but earning a dollar yourself is still outside the norm for most people — "I had an idea, and someone paid for it." He says it's like opening a small door, and you become more ambitious, going from twenty or thirty dollars to tens of millions in revenue. He also admits you might try it and find you don't like it, but everyone should try it once, and most people are better at it than they think.
— James CliftIn their own words · checked verbatim
We want to make owning a business easier than having a job.
James Clift0:00
now all of a sudden you're going from making 20 bucks an hour to 100 bucks an hour or 200 bucks an hour for the exact same work
James Clift5:10
business is thousands of tasks that are so important, but actually. not valuable
James Clift19:30
no one wants to click on a bunch of stuff anymore
James Clift32:54
you have to assume that the models will keep getting better and better and will will make your product obsolete it probably already is obsolete but like at least like in the next six months it will be again
James Clift36:01
I think for me it was actually the, the solo founding took a lot of that pressure off
James Clift47:28
the most interesting part about just making a dollar that you've made on your own is it's so outside the norm for most people still
James Clift1:03:49
Figures
| Durable registered users | about 3.5 million | 10:19 |
| Share of Durable registered users that are net new | more than half | 10:19 |
| Number of Durable engineers | 6 | 31:50 |
| Visual CV revenue after a year and a half | about a few million dollars | 39:10 |
Glossary
- GEO
- Generative engine optimization: optimizing content so it gets cited and recommended by generative engines like ChatGPT.
- GMV
- Gross merchandise value: the total value of transactions on a platform, often used to measure market size.
- South Park Commons
- A fund and startup community in San Francisco that invests in founders at the "minus one to zero" stage.
- vibe coding
- A development approach where you use natural language to have AI generate code directly and ship products fast.
How to listen
Founders considering going solo or already solo, and product leads building AI products who are torn about whether to keep building UI.
The industry chit-chat and Shopify analogy from 0:00-9:00 can be fast-forwarded.