What this project is

fly.pieter.com is a browser-based free-to-play MMO flight simulator. You land, pick a plane, fly and dogfight other players online. No registration, no downloads.

Pieter Levels built the prototype in February 2025 in 3 hours using Cursor, Claude, and Grok 3. Seventeen days later the project was making $87,000 MRR ($1M ARR). A year on, it sits around $70-90K MRR — on one person, no team.

This is not breaking news — the case is from March 2025. But it became the reference point for the entire vibe-coding community: "here is how a solo with AI made a million in two weeks." A year later, let's separate signal from noise.

How it makes money

Three channels, in revenue order.

1. In-game sponsor slots — the main income. Planes with brand logos fly around. Billboard-planets hover. Floating spheres link to sponsor sites. This isn't AdSense — these are direct prepaid contracts. A brand pays a fixed amount for a month or two, their logo becomes part of the world.

The main buyers are AI startups looking for developer and tech-enthusiast audiences. Pieter built no sales team. He sold slots directly via tweets: "3 ad slots left, DM me." The queue formed itself.

2. Custom brand integrations. The famous example: Basecamp ordered a custom UFO carrying their signature snow-globe logo. Not a banner — the actual in-game object, designed around the brand. Costs more than a regular slot.

3. In-game purchases. F-16 for $29.99 and other upgrades. Small revenue, but matters as validation: people are willing to pay inside a browser free-to-play game, which is not a given.

How he promoted it

This is the part everyone skips when they cheer.

The base — 600,000 Twitter followers, built over 10 years. Over that decade Pieter launched 40+ products, posted daily, openly shared every number — revenue, costs, screwups. When he launched fly, a crowd was already standing there. Without that base, nothing else works.

Day 0: a tweet saying "3 hours of code, here's the video." A short gameplay GIF, not a wall of text. Video always beats screenshots.

Days 1-3: daily updates. New feature = tweet with GIF. New plane = tweet. Every number = tweet. Speed itself generates content — you don't need to "think of what to post."

Around day 3: Elon Musk retweeted with "Wow, this is cool. AI gaming will be massive." 200M+ followers saw it. That was the moment "going viral" turned into "already viral."

Days 5-17: revenue updates every 2-3 days. "$10K MRR," "$30K," "$60K," "$87K = $1M ARR." Each number is its own tweet with a Stripe screenshot. The graph becomes content — people retweet the story, not the game.

In parallel: selling sponsor slots straight from tweets. When you have 320K players and an arrow going up, brands come to you.

What's actually replicable

Honestly — not much. But there is something.

1. Build in public from day one. Even with 100 followers — that's already an asset. In a year you might have 1000. In three years, 10K. Without that start, you're in the same spot a year from now.
2. Public metrics as content. The cheapest, most reliable content format for a solo founder. You don't have to "think of what to post" — every number is already a story.
3. Video demos over descriptions. Works on any account, even a brand-new one. A short GIF of the product in action beats any text.
4. Controversial positioning. A strong claim ("AI gaming is the future") attracts argument. Arguments get more impressions than neutral posts.
5. Find your niche community. Pieter had Twitter. You might have Reddit, Discord, a specialized Telegram chat. The key is a place where you already have credibility.
6. Sell slots, not subscriptions. If you have an audience, sell them placement next to you. It works faster than building a SaaS funnel.

What's NOT replicable

Let's name this directly.

• The Musk retweet. That doesn't repeat — it's a lottery ticket.
• A 10-year reputation in tech community.
• The network of other founders who repost and amplify.
• Access to brands willing to buy ad slots over DM, no contracts, no procurement.

Strip all this away, and what's left is a product built in 3 hours that, without distribution, would have made $1K MRR instead of $1M ARR.

The main takeaway

"$1M ARR in 17 days" is not a viral launch — it's the conversion of a 10-year audience asset. The product was the trigger, not the engine. The engine was distribution, which Pieter spent a decade and a half building.

For a solo founder looking at this case, the lesson isn't "how to build a game" — it's "how to start building your audience starting today." In five years, you'll have your own "17-day launch." Without those five years, you won't.

What to do this week

• Open Twitter/X (or LinkedIn, or Reddit — wherever you already have something) and post what you're building. Not "when it's ready." Today.
• Make it a rule: one post a day about the process. A number, a screenshot, a small update, a failure, a lesson. Anything. The key is every day.
• Make a 30-second GIF of your product in action. If you don't have a product yet — a GIF of any part that works.
• Follow 20-30 people in your niche who build in public. Not for networking — for the pattern. Watch which posts of theirs land.
• Build the habit of showing numbers. Any of them. Even when they're still small.

Distribution is the long game. The product is the short sprint at the end of it. Pieter Levels is living proof.