Case study

How Pieter Levels Runs a $3M/Year Company With Zero Employees

Before AI agents existed, Levels proved one person with taste and stamina could outship 50-person startups.

Pieter Levels built 70+ projects, 3 of them past $38K MRR each, solo since 2014. The original one-person company playbook.

Free to startNo credit card requiredUpdated Apr 2026
Revenue
$3M+ annual run-rate, ~87% net margin
Employees
0 — solo since 2014
Industry
Multi-product indie SaaS
Founder
Pieter Levels

Timeline

2014
Launches 12 startups in 12 months as a public challenge. Nomad List emerges as the breakout — a community ranking cities by remote-work quality.
2015-2017
Ships Remote OK (jobs board), Hood Maps (neighborhood vibes), and dozens of smaller experiments. Builds everything in Vanilla PHP, jQuery, SQLite.
2018-2022
Nomad List crosses $30K MRR. Remote OK becomes the most-trafficked remote jobs board in the world. Levels refuses to hire, arguing small + focused beats big + distracted.
2023
Launches Photo AI and Interior AI on the back of Stable Diffusion. Photo AI hits $100K+ MRR within months. Confirms that a solo founder with distribution can ride every wave faster than VC-backed teams.
2023-2024
Photo AI scales from $5.4K in week 1 → $28.7K in month 2 → $61.8K by month 6 → crosses $100K MRR around month 18.
March 2025
Launches a flight simulator experiment that hits $1M ARR in 17 days (publicly reported on X at $87K MRR) — proving the portfolio model still works at 2025 velocity.
November 2025
Photo AI reported at $132-138K MRR (roughly $1.65M ARR, ~70% of his total income). Remote OK at ~$41K/mo. Interior AI at ~$40K/mo. Cost base is ~$13K/mo, mostly Replicate GPU bills.
2026
Continues shipping experiments weekly. Runs the entire portfolio from a laptop, mostly while traveling. Outsources almost nothing except customer support for high-volume products.

Key insights

  • 01Portfolio beats perfection. Levels runs 70+ projects because most will fail and a few will compound. The cheap ones keep the lights on; the winners pay for the next decade.
  • 02Ship in public. Posting MRR, revenue screenshots, and honest postmortems on X is not marketing — it is the flywheel that funds every new launch.
  • 03Boring stack, relentless velocity. Vanilla PHP + jQuery + SQLite is unfashionable and devastatingly effective for a single person who knows it cold.
  • 04Community compounds. Nomad List's paying members are a moat that no GPT can replicate — the relationships between members are the product.
  • 05Margins over growth. Levels refuses VC, refuses hires, and refuses scaling at the expense of margin. Net result: 90%+ profit margin, full ownership, asymmetric freedom.
  • 06AI did not invent the one-person company — it accelerated a pattern Levels has run for a decade.
  • 07Taste is the bottleneck. Levels ships faster than anyone because he has opinions about what to build and the confidence to kill things fast.

Stack used

Vanilla PHP (no framework) on a single VPSjQuery for front-end interactivitySQLite as the primary database for most projectsCloudflare in front of everything for caching and DDoS protectionStripe for payments and subscriptionsStable Diffusion / Replicate for Photo AI and Interior AIX/Twitter as the primary growth channelPlausible Analytics for privacy-friendly trackingEmail providers rotated based on deliverabilityDiscord + a custom in-product forum for community products

What this means for you

  • Ship more, smaller. A portfolio of 10 tiny products teaches you more than one over-engineered startup.
  • Pick a stack you can hold in your head. The productivity gain of a simple stack outweighs any framework's marketing.
  • Build an audience before you need it. Levels' 700K+ followers are not a vanity metric — they are the distribution moat every new launch inherits.
  • Refuse to hire until you are literally unable to ship the next thing. Most founders hire to outsource discomfort, not to scale output.
  • Use AI to raise your ceiling, not to change your identity. Levels stayed Levels; AI just meant he could ship Photo AI in weeks instead of quarters.
  • Optimize for freedom. A one-person, high-margin, portfolio-of-products company is a better life than a $50M Series B where you are a middle manager.

Frequently asked questions

Is Pieter Levels really solo, or does he have hidden contractors?

Levels has publicly stated many times that he runs his businesses alone, including payroll, customer support, and engineering. He occasionally uses contractors for localized tasks (translations, some customer service spikes) and partners with companies like Replicate for model inference. But there is no hidden team of engineers. His companies are unusual precisely because one person, one laptop, and a boring stack are genuinely doing all the work. This is why he has become the founder figure everyone points to when the 'one-person company' conversation starts.

Why does Levels refuse to hire even at $3M/year?

Two reasons he has explained publicly. First, management is its own job: adding even one person changes his work from building to explaining, reviewing, and negotiating. He does not want that job. Second, margins: each hire consumes the optionality that lets him kill projects, pivot, or take a month off. At 90%+ profit margin, the marginal cost of not hiring is negligible; the marginal upside of hiring is mostly imaginary. For Levels, staying solo is not a constraint — it is the product.

What does Levels' stack reveal about modern best practices?

It reveals that 'best practices' in consumer SaaS are often a function of team size, not product success. A 50-engineer company needs TypeScript, microservices, and CI pipelines because coordination is expensive. A solo founder has zero coordination cost, so the best stack is whichever one they can hold in their head and ship with fastest. Levels using Vanilla PHP and SQLite in 2026 is not a nostalgic choice — it is a productivity optimum for his specific constraints. Your stack should look like your team shape, not like Google's.

Can someone without a 700K-follower Twitter account replicate Levels' playbook?

Yes, but the path is slower. Levels' distribution advantage is the compounding result of a decade of public building. The replicable parts are: pick a real pain, ship something in weeks not months, post the journey publicly, charge money immediately, kill things that do not work, and keep a portfolio. An audience in 2026 can be built on X, LinkedIn, YouTube, or a niche newsletter — the specific channel matters less than the discipline of showing up with real work. The one thing you cannot shortcut is time.

What would Pieter Levels build if he started over in 2026?

He has hinted at this in interviews and on X. The answer tends to be: pick a vertical where AI creates a new wedge, ship something small in 2-4 weeks, charge for it from day one, and compound. Concretely, he often points to AI-powered creator tools (like Photo AI), niche vertical SaaS (where incumbents are slow), and community products where AI can remove the tedious parts of moderation. The pattern is the same as before — find a demand signal, build fast, charge immediately — only the toolchain has changed.

Related resources

Run your one-person company.

Hire your AI team in 30 seconds. Start for free.

Free to start · No credit card required · Set up in 30 seconds