Pieter Levels: How He Built a One-Person Business

by TFE Editor · 12 min read
Pieter Levels: How He Built a One-Person Business

Pieter Levels, known online as levelsio, is a Dutch indie developer who builds and runs internet products mostly by himself, with no venture funding. His best-known sites are Nomads.com (formerly Nomad List), Remote OK and Photo AI. For a solopreneur he is the clearest example of a high-margin, tiny-team business, and of how many attempts it takes to find the winners.

If the word is new to you, start with our guide on what a solopreneur is. This profile goes much deeper on one person: what he built and when, the numbers he has published (dated and attributed, because they are self-reported), where his approach has limits, and how you could run the same play today with AI doing the first drafts. It is part of our series on solo founders.

Pieter Levels at a glance

Item What the sources say
Known for The “12 startups in 12 months” challenge (started 1 March 2014); Nomads.com (formerly Nomad List); Remote OK; Photo AI; the self-published book MAKE
Business model Subscriptions and one-off payments for web products; paid job posts on Remote OK; book sales. He says he does not use venture funding.
Team His blog says he builds the startups “by myself.” In a 2024 interview he said he had once hired a developer for about ten months, for Python-based AI work on Photo AI.
Stack PHP, jQuery and SQLite, by his own account (2024 interview). He hosts on his own servers.
Self-reported scale Business revenue of about $200,000 to $250,000 a month as of 23 September 2026, with a profit margin he put at 94.5%
Where to follow His blog at levels.io, his X account @levelsio (about 980,000 followers on his stats page, October 2026) and his email newsletter (his homepage says about 185,000 people read his posts in full by email, October 2026)

The story: from a music channel to a portfolio of small products

Before the startups (to 2013)

Levels has said he graduated university in the Netherlands and did not want a conventional job. He was earning money from a YouTube music channel; in the 2024 Lex Fridman interview he put the AdSense income at about $2,000 a month at one stage. He then moved to Asia and began working remotely. His own post “Reset your life”, dated 22 April 2013, describes selling his belongings and leaving with a laptop and a backpack. His projects page says he ended up staying in Asia for over a decade; in the Fridman interview he also describes a low stretch that included a return to Holland, so treat the timeline as his own summary.

March 2014: the challenge that started it

On 1 March 2014 he published “I’m Launching 12 Startups in 12 Months”. The rules were simple: one product a month, from idea to launch, with each project given a month to show signs of demand. His stated reason was his own habit of not finishing or launching things. In his words: “I want to change my habits and force myself to finish what I start.”

In the Fridman interview he added that his income had fallen to roughly $500 a month by then. The challenge was his way of turning that around by shipping more.

The list of launches on his site includes Go Fucking Do It (a goal tool that charges your card if you miss your deadline), Tubelytics (a YouTube analytics dashboard), Nomad List, Nomad Jobs, GIF Book and the #nomads chat community. Remote OK appears on the same list, but his projects page dates its launch to February 2015.

Nomad List and Remote OK (2014 to 2021)

Nomad List began, he told Fridman, because he was in Chiang Mai, Thailand, living on about $500 a month, and wanted to find cities with fast internet and cheap rent. He has said it launched as a public Google spreadsheet anyone could edit, and that he also built it because he felt lonely; he told Fridman it grew into a community of about 30,000 to 40,000 members. He put his rent for a private room there at about $150 a month. His projects page dates Nomad List to July 2014, and it has since been renamed Nomads.com.

Remote OK came out of the job board on Nomad List: companies paid for job posts, so he spun it off. A few benchmarks he posted along the way:

  • 5 January 2016: his 2015 revenue was $202,785, with one server, one laptop, no office and $0 funding.
  • 30 May 2019: Nomad List and Remote OK reached $83,333 a month in revenue, which he noted multiplies out to $1 million a year. He listed 11,996 customers, one server and no ad budget.
  • 12 April 2021: Remote OK generated $101,101 in a month, which included about $9,500 of ads, by his own breakdown.

The book (2018)

His projects page dates the self-published book MAKE, about building startups without funding, to March 2018. As of 23 October 2026 the book’s website shows a counter of 32,173 copies and $956,108 sold. That is a live counter on his own site, not an audited figure. The book’s introduction warns of survivorship bias and says plainly: “No, this book won’t make you successful.”

The AI products (2022 to 2023)

In 2022 he moved into AI. In the Fridman interview he traced a lineage from a site called This House Does Not Exist to Interior AI, then Avatar AI, then Photo AI. His post of 4 October 2022 says he built Interior AI in five days. A few weeks later, on 7 November 2022, he reported selling $100,000 of AI-generated avatars in 10 days through AvatarAI.me: 2,943 sales at an average of about $33. On 3 July 2023 he said Photo AI, the descendant of those experiments, had 1,872 paying customers and $61,808 a month in revenue. In the Fridman interview he summed up the pitch as “fire your photographer.”

2024 to 2026: bigger audience, bigger numbers

  • 20 August 2024: Lex Fridman Podcast episode #440 with Levels, which put his approach in front of a mainstream audience.
  • 22 February 2025: in a post dated 22 February 2025 he said the browser flight simulator fly.pieter.com, built mostly live on X, reached $1 million in annual recurring revenue in 17 days (his claim).
  • 6 March 2026: he posted that Photo AI made $105,000 a month in revenue and $80,000 a month in profit.
  • 23 September 2026: he posted that he had passed $10 million a year in revenue plus investment gains, at a 94.5% profit margin, and that business revenue had stayed stable at about $200,000 to $250,000 a month. Most of the investment gains, he said, were unrealized.

Annualized, $200,000 to $250,000 a month is roughly $2.4 million to $3 million a year (our arithmetic). That fits a July 2025 Techleap profile, which described his businesses as generating over $3 million in annual revenue. Treat all of these as his own unaudited figures.

The playbook: five principles

1. Ship something small, in public, on a deadline

The 12-month challenge worked because it was a public promise with a short clock. In the Fridman interview he described launching a minimal version within about two weeks and checking for demand, and said publishing the plan created accountability. The product did not have to be good. It had to exist, so reality could answer.

2. Let money be the test

He told Fridman that signups are not enough: people need to take out a credit card. That is why so many of his early launches included a payment button. A user count tells you what people tolerate. A payment tells you what they value. This applies equally if you sell a service, which is why a productized service with a fixed price is the non-coder’s version of the same test.

3. Keep the stack and the costs boring

He builds with plain PHP, jQuery and SQLite and runs his sites on his own servers. On 18 April 2026 he posted that his sites handle about 4 billion requests a year for about $2,932 a year in hosting. The point is not that you should use PHP. It is that a stack you already know, with low fixed costs, lets you try many ideas cheaply and keep most of every dollar.

4. Let winners grow out of earlier projects

Remote OK began as a page on Nomad List. Interior AI came out of a house-generation experiment. Photo AI came out of avatar and interior-design work. Each step reused skills, an audience or a working feature from something before it. Few of his winners started from a blank page.

5. Automate, and protect the margin

He told Fridman his companies run at around 90% profit margins and are highly automated, and that he tries not to hire. His 23 September 2026 post describes a deliberate push to cut software subscriptions by building his own replacements. The lesson for a solopreneur is the same one we make in our piece on solopreneur vs entrepreneur: a small business is judged on what it keeps, not on what it grosses.

What didn’t work, and the caveats

  • Most projects fail, by his own count. His project list (updated 3 October 2026) tags 116 entries: 9 as successes, 11 as “okay”, 19 as failed, 71 as non-profit or not aimed at profit, and 6 as too new to judge. In a November 2021 post he put his hit rate at about 5% and added: “So…ship more.” The shipping rate is part of the strategy, not a footnote.
  • Some hits didn’t last, or earned little. He tags the flight game and the avatar app as “okay: took off, made money but not sustainable”, the same label he gives Rebase, an immigration-as-a-service project. He also told Fridman that Hoodmaps never made money and once produced a $20,000 hosting bill, which the provider helped reduce.
  • Revenue swings. On 22 September 2024 he posted that he had hit a new $420,000-a-month revenue record thanks to the Lex Fridman podcast (headline on his homepage; self-reported). He also told Fridman that Remote OK peaked at about $140,000 a month in 2021, fell to about $10,000, and was, he thought, back to about $40,000 (his hedged estimate, in 2024). If you are building on a market boom, plan for a bust. Our piece on surviving uncertain economic times covers the cash side.
  • Survivorship bias. He raises this himself in MAKE. We see the winners because he posts them.
  • An audience head start. A March 2025 404 Media article on the flight game noted that Levels already had a large following and a record of launches, which not every new builder has. The same article carried a headline claim of $50,000 a month and a subhead saying “thousands of dollars a month”, so the sources disagree on that number and we leave it out of the timeline.
  • “Solo” has asterisks. The site says he builds by himself, but he told Fridman he once hired an AI developer for about ten months and was building a new AI startup with a close friend who writes the code (as of 2024). Solo means he keeps control, not that no one else ever helps.
  • Self-reported numbers. Almost every figure above comes from his own posts or interviews. We have not seen audited accounts.

The AI version of this playbook

This section is our analysis, not Levels’s advice, and the tools change quickly, so it names none. The sequence is his: small, public, paid, repeat. The difference is who does the first draft.

Step What you decide What AI drafts What you check
Pick the problem Whose annoyance, and whether they would pay Lists of existing alternatives, search-demand summaries, ten ways to phrase the offer Talk to a few real buyers. AI summaries of demand can be wrong.
Build the smallest version The one feature that matters; what to leave out A prototype, landing-page copy, a payment page Run a real payment end to end; review security and anything that stores user data
Launch in public Where your buyers already gather Launch posts, a short demo script, reply drafts Every claim and number; that it sounds like you
Run it Prices, refund policy, what you refuse to do First replies to common questions, help pages, weekly usage summaries Anything involving money, refunds or complaints
Keep or kill The kill rule, set before launch A summary of the numbers against your rule That you are reading the numbers honestly

The checking column matters more than the drafting column. Our pieces on getting it right, not just fast and training AI tools to work your way go into how.

A first-week plan (our suggestion)

  1. Day 1: List ten irritations from your own work or life. Pick the one you would pay to fix.
  2. Day 2: Write the offer in one sentence, a price, and a kill rule, for example “if nobody has paid by day 14, I stop.” Browse our one-person business ideas if the list is thin.
  3. Days 3 and 4: Build the smallest version that could take payment. Let AI draft the page and the first working version. You make the scope decisions.
  4. Day 5: Put a real payment link live and test it with your own card.
  5. Day 6: Launch in one place where your buyers are, with an honest post about what it does and does not do.
  6. Day 7: Review. Count payments, not visits. Keep, change or kill, as you promised on day 2.

Lessons for your one-person business

  • Set a short deadline and announce it. A public promise does work that willpower won’t.
  • Count payments, not signups or compliments.
  • Expect most attempts to fail and make each one cheap enough that this is fine.
  • Use what you already know. A plain, familiar stack beats a fashionable one you have to learn first.
  • Build the next product from the last one: reuse the audience, the code and the customers.
  • Guard your margin. Every subscription and hire has to earn its place.
  • Don’t extrapolate from a viral spike. His own “okay” tags show how that goes.
  • Decide what kind of business you want first. Levels’s route is one version of a lifestyle business; plenty of solo owners choose slower, steadier versions.

Pieter Levels FAQ

How much does Pieter Levels make?

He has said that as of 23 September 2026 his business revenue was stable at about $200,000 to $250,000 a month, and that he had passed $10 million a year in revenue plus investment gains with a 94.5% profit margin. In March 2026 he posted that Photo AI alone made $105,000 a month in revenue and $80,000 in profit. These are his own figures and have not been audited.

Does Pieter Levels have employees?

His blog says he builds his startups by himself. On his homepage he describes himself as building his startups by himself, bootstrapped, with him coding everything on his laptop. In the 2024 Fridman interview he said he doesn’t hire people, but also that he hired a developer for roughly ten months to help with Photo AI. We have no source for his current staffing beyond his own statements.

What is the “12 startups in 12 months” challenge?

It was a public challenge he began on 1 March 2014: launch one new product every month for a year. Two of the products, Nomad List and Remote OK, became lasting businesses, according to his own project list.

What tech stack does Pieter Levels use?

He said in 2024 that it is PHP, jQuery and SQLite. He has also described using AI tools for some projects, such as the flight game.

Can I copy his approach?

You can copy the method: small launches, paid tests, low costs, frequent attempts. You cannot copy his head start, which includes an audience built since 2013, and his own numbers show most attempts fail. Treat his results as an upper range, not a forecast.

The takeaway

Levels did not find one brilliant idea. He ran a cheap, repeatable process many times, kept his costs near zero and let the few winners compound. The method transfers to a one-person business with AI as your team; the results will not be guaranteed, and his own data says most tries will fail. Make the tries cheap and the kill rule firm. If you want the practical version, grab the free “One-Person Business AI Stack” checklist: every business function, the workflow to hand to AI, and what to keep for yourself.

Sources

All figures are self-reported by Levels unless stated. Pages were read in October 2026.