Peter Steinberger sold his company for over $100 million. Then he fell apart.
After 13 years of building PSPDFKit, a PDF software used by Apple, Disney, and Dropbox.
He sold his shares, fell into a hole, and spent three years doing therapy, travelling, and figuring out what the point of anything was.
Then he came back and started building again.
He built 43 projects. Every single one went nowhere.
Project 44 was a Saturday night hack he put together in about an hour.
He glued WhatsApp to an AI model so it could actually do things for him, book flights, clear his inbox, run tasks in the background while he slept. He posted it online mostly just to share it.
It hit 9,000 GitHub stars in 24 hours. Then 60,000 in two weeks. Then 100,000 in 48 hours after it went properly viral.
Even Zuck texted him personally. Both Meta and OpenAI made offers implying billion-dollar valuations.
Let that sink in…
One guy with no team, funding, or a marketing budget.
Now here's the part of this story that actually matters.
The whole time OpenClaw was going viral, it was costing Peter $10,000 to $20,000 a month out of his own pocket to keep it running.
And he was pumping every dollar of sponsorship money to the other open-source developers whose tools he'd built on top of, keeping absolutely nothing for himself.
He wasn't just in it for money. He was just building something people actually needed.
And that's the first lesson worth taking from this.
He didn't write a blog post about what AI agents could do. He didn't make a course.
He built the actual thing, made it free, made it open source, and let people run it themselves.
His users became his marketers. Word of mouth did what no ad budget ever could.
Which brings us to the 43 projects.
People love to skip over that part. They see the viral moment and assume it was inevitable. It wasn't.
Peter built OpenClaw out of burnout and a personal frustration.
He'd been failing for years before that one Saturday night.
Most people quit at project three because they want proof before they put in the work.
Peter kept building until the work created its own proof.
43 failed projects before the one that made major tech CEOs pick up the phone.
When both Meta and OpenAI came calling, Peter turned down building a standalone company.
His exact words:
"I could totally see how OpenClaw could become a huge company. And no, it's not really exciting for me."
The man had already built a $100M company. He knew what that game looked like. He chose to keep building instead…
He joined OpenAI on Valentine's Day, and stated mission: build an agent his mum can actually use.
Simple as that.
So what's the actual takeaway here?
Think about the thing in your business that you hate doing every single week.
The repetitive stuff
The back and forth emails
The research rabbit holes
The content reformatting.
That's your version of the problem Peter solved.
And there's probably already an AI tool, or a simple workflow you could build in an afternoon, that handles it.
The other thing is the 43 failures. We talk a lot about execution, but we don't talk enough about volume.
Peter didn't get lucky on project 44. He got lucky because he built 43 things before it. Luck finds the people who are still in the game.
And the last thing… he made it free. He built trust before he built revenue.
That's the same reason content works for your brand.
You give first, consistently. And when you eventually ask for something, people actually listen.
The window for building something real with AI has never been more open.
One person with a clear problem and the willingness to ship fast just made every major tech CEO pick up the phone.
You've got the same tools he had.
Shut up and execute.
