The story.
01 — How it got builtPieter Levels was born in 1987 in the Netherlands. He has no computer science degree. He studied music production at the Conservatorium van Amsterdam, worked as a techno DJ, and uploaded music tutorials to YouTube, growing to 139K subscribers. He learned to code in his late twenties by building real projects and Googling solutions on Stack Overflow.
In March 2014, Levels made a public commitment: launch 12 startups in 12 months. One new product every month. The constraint forced him to focus on core functionality and ship fast. His argument: a startup doesn't have to be a world-changing billion-dollar company from day one. AirBnB started selling Obama-themed cereal. Dropbox was just a GUI for rsync.
Most of the 12 startups failed. Play My Inbox (friends send music over email) didn't make money. GifBook (animated GIF flip books) had 2-3% margins. Go F*cking Do It made $500/month and flatlined. But startup #4 changed everything: Nomad List.
5% hit rateOnly 4 out of 70+ projects ever made money. 95% of everything Pieter built failed.
The idea came from his own problem. While traveling and working remotely, he had no way to compare cities for remote work. He created a Google Sheet, tweeted it out, and accidentally set permissions to "anyone can edit." Strangers started adding cities and data. It went viral on Reddit and Hacker News. He built a proper MVP in PHP within a month.
Nomad List kept growing. He started charging $5 to access the community chat (to combat spammers), then raised to $25, then $50, then $65. Sign-ups stayed the same or grew at each increase. He launched Remote OK in 2015, a remote job board that now does $3.4M/year. Every filter on Nomad List generates a unique URL, creating hundreds of long-tail SEO pages automatically. Customer acquisition cost: $0.
In late 2022, Levels started experimenting with Stable Diffusion for AI image generation. He built Interior AI (room redesign tool), which hit $10K in its first week. Then he noticed fine-tuned models could generate photorealistic faces. He uploaded his own photos, was amazed by the results, shared them on Twitter, and launched AvatarAI the next day. It made over $100,000 in its first 10 days. This evolved into Photo AI, launched February 2023.
$100K in 10 daysAvatarAI launched from a Twitter post. Evolved into Photo AI, now his biggest product.
Photo AI became his biggest product. $61K MRR by July 2023. $100K MRR by September 2024. Currently $130K+/month with 87% profit margins. GPU costs are only $13K/month. The product generates roughly 70% of his total income.
The entire empire runs on PHP, jQuery, and SQLite. No React, no Next.js, no Docker, no Kubernetes. Each app is primarily a single index.php file on a single VPS. His reasoning: "PHP just stays the same and works. People are getting sick of frameworks." He has zero employees. No office. He lives out of a backpack, works from coffee shops in 40+ countries, and runs everything from a laptop. Total portfolio: $3M+ ARR across Photo AI, Interior AI, Nomad List, and Remote OK. His own assessment: "Only 4 out of 70+ projects I ever did made money. My hit rate is only about 5%."
0 employeesOne person, a backpack, PHP. $3M+ ARR across four products. No office, no team, no meetings.
95% of everything he built failed. Out of 70+ projects, only 4 made money. But those 4 generate $3M+/year, run by one person, on PHP and jQuery, from a backpack.