I don't know how to code. I built a SaaS product anyway. Here's every number, every mistake, and the moment I almost shut it down.

The first line of code I ever wrote — or rather, the first line of code an AI ever wrote for me — was in November 2024, at 2am, in my kitchen in Berlin. I was standing because I couldn't sit still. The nervousness was physical, concentrated in my hands, which were hovering over a keyboard like someone approaching a wild animal. The screen showed a terminal window. The terminal showed a cursor. The cursor blinked, and I blinked back at it, and I thought: What the hell am I doing?

I was building a SaaS product. Me — a Korean woman with no technical background, no CS degree (no degree at all), no engineering friends, no CTO, no technical co-founder. I had an idea and a Claude subscription and the specific flavor of delusion that comes from having already built one impossible thing and starting to believe the impossibility is the method.

Soulin Social started because I was drowning in my own content. KINS needed social media. The blog needed distribution. The newsletter needed scheduling. I was spending fifteen hours a week on content operations — writing, scheduling, reformatting, posting, cross-posting — and every hour spent on content was an hour not spent on the thing that actually mattered: the healing work, the protocols, the guests.

I looked for tools. I tried them all — the schedulers, the AI writers, the content calendars. Each one solved a piece of the problem and created three new ones. They were built for marketers, for teams, for people whose entire job was content. I wasn't a marketer. I was a solo founder who needed content to happen without it eating my life.

So I decided to build the tool I needed. The fact that I couldn't code was, in my mind, a minor obstacle. I'd built a hotel without a hospitality degree. How much harder could software be?

The answer, it turns out, is: a lot harder. But also not in the ways I expected.

Let me give you the current numbers first, and then I'll tell you how I got here. Because the numbers without the story are meaningless, and the story without the numbers is just another founder narrative.

Soulin Social, March 2026:

• Monthly recurring revenue: 2,900 euros
• Total active users: 187
• Paying users: 94
• Free tier users: 93
• Monthly churn: 8.2%
• Average revenue per paying user: 30.85 euros

Those numbers are small. I know they're small. In the SaaS world, these are seed-stage numbers, pre-product-market-fit numbers, the kind of numbers that would make a VC close their laptop and check their phone. I'm sharing them because they're real, and because the gap between "real SaaS numbers from a solo non-technical founder" and "the SaaS numbers you see on Twitter" is a canyon filled with survivorship bias and lies.

The building process was the most humbling experience of my professional life. And I built a hotel alone, so the bar for humbling was already high.

The first version of Soulin Social took three months to build. Three months of nightly sessions with AI — describing what I wanted, receiving code, trying to understand the code, breaking the code, describing the breakage, receiving new code. I didn't learn to code. I learned to communicate with something that codes. The distinction matters. I still can't open a blank file and write a function. But I can describe, with increasing precision, what I need a function to do, debug the output, and guide the iteration.

The first version was terrible. I shipped it anyway. Twelve people used it — all friends, all obligated by loyalty to try it, all too polite to tell me it was bad. But their usage data told me. Average session time: four minutes. Feature adoption: 23%. The tool I'd built to solve my own problem was too confusing for anyone else.

I rebuilt it. Not from scratch — I couldn't afford the time or the emotional cost of starting over. I rebuilt it the way you renovate a building while people are still living in it: one room at a time, hoping the ceiling holds.

Version two was better. I simplified the interface by removing 60% of the features. This felt like cutting off limbs. Every feature I'd built represented hours of work, nights of debugging, the specific pride of a non-coder who made something function. Deleting them was painful in a way that was disproportionate to the act. But the usage data was clear: people used three features. Everything else was furniture — it made the room look full but nobody sat on it.

The growth came in stages, each one teaching me something I didn't want to learn.

Stage one: Friends and pity users (months 1-3). Twelve users. Zero revenue. I was too afraid to charge. Charging meant the product had to be worth money, and I wasn't sure it was. The fear of charging was, I realized later, the fear of being judged — the Korean good-girl conditioning that says receiving money for your work means you're being presumptuous about your own value.

Stage two: First strangers (months 4-6). I wrote a blog post about building the tool. It ranked for a long-tail keyword I hadn't targeted. Twenty-three people signed up in two weeks. They weren't friends. They had no obligation to be kind. Their feedback was direct and sometimes brutal. One user emailed me: "This is almost good. The scheduling works but the AI suggestions are generic. I can get generic from ChatGPT for free." My face burned reading it. Then I fixed the AI suggestions. That email improved the product more than three months of building in isolation.

Stage three: First revenue (month 7). I launched a paid tier at 29 euros per month. Eleven people converted from free. 319 euros per month. I remember staring at that number and my body doing two things simultaneously: my chest swelled with something that might have been pride, and my stomach clenched with the certainty that it would disappear. The scarcity brain doesn't celebrate income. It preemptively mourns its loss.

Stage four: The almost-quit (month 10). Revenue had grown to about 900 euros per month. But the support load was crushing me. Every bug report came directly to my inbox. Every feature request. Every "how do I..." question. I was spending more time supporting the product than building it. The KINS work was suffering. My writing was suffering. I was sleeping less. And the revenue — 900 euros per month — was not enough to justify the cost to my health and my primary business.

Soulin members get the full essay library, private group chat, the Soulin OS e-book, and every tool — all for $10/mo. Join Soulin →

Full essay library · Private group chat · Soulin OS e-book · Every tool · $10/mo

I drafted a shutdown email. I sat with it for three days. On the third day, a user — someone I'd never met, never spoken to — posted in the community forum: "Soulin Social saved me four hours a week. I used those hours to launch my podcast." And I closed the shutdown email and opened the bug tracker.

I'm not proud of how close I came. But I'm honest about it because the "almost quit" moment is in every founder's story and nobody writes it down.

What broke (a partial list):

The payment integration failed silently for eleven days in month five. Eleven days where new users were signing up for the paid tier and not being charged. I didn't notice because I was managing a KINS crisis. When I finally checked, I had forty-three users who'd been using paid features for free. I couldn't retroactively charge them. I ate the loss — roughly 1,200 euros — and fixed the integration and cried in the bathroom. Not from the money. From the exhaustion of being the only person watching the machine.

The AI content suggestions began hallucinating brand voices after an API update. Users were getting output that sounded nothing like their established tone. Three users churned that week. I spent forty hours — across five days — rewriting the prompt architecture. Forty hours I didn't have, stolen from sleep and KINS and my body, which registered the theft as a persistent tremor in my right hand that lasted two weeks.

The database hit its limit in month eight and the app went down for six hours on a Tuesday. I learned about database scaling the way I learn everything: in crisis, at speed, with my heart rate elevated. The fix took fourteen hours of continuous work. I ate two granola bars and drank cold coffee and when it was finally working again, at 4am, I lay on the kitchen floor and stared at the ceiling and felt the specific loneliness of being the only person who would have noticed if the fix hadn't worked.

What worked:

The "voice learning" feature — where the AI analyzes a user's existing content and learns their tone. This was the feature that made people stay. The churn rate among users who activated voice learning was 3%. Among those who didn't: 14%. One feature. The difference between a product people tolerate and a product people need.

The weekly email digest — an automated summary of what the tool scheduled, what performed well, what could be improved. Users told me this was the first thing they read on Monday morning. It cost me almost nothing to build and created more perceived value than features that took weeks.

The community forum, which I almost didn't build because "community" felt like a buzzword. It turned out that solo founders using an AI content tool were hungry for connection with other solo founders using an AI content tool. The forum became a support channel that reduced my direct support load by 60%. Users answered each other's questions. The product became a place, not just a tool.

Where it is now:

2,900 euros per month. 187 users. 8.2% churn that I'm working to bring below 6%. A product that runs — mostly — without my constant attention, thanks to the AI agents I've built for monitoring and first-line support. I spend about ten hours a week on Soulin Social. Some weeks more, when things break. Some weeks less, when everything holds.

It's not a success story yet. It might never be, by SaaS standards. It might stay small — a tool for a specific kind of person, a solo founder who needs content to happen without it taking over their life. That's who I built it for. That's who uses it. And the revenue, while small, is real and growing and mine.

The girl at the kitchen terminal at 2am, hands hovering, cursor blinking — she'd be shocked by 2,900 euros per month. She'd also be shocked by the bathroom crying and the database crashes and the forty-hour prompt rewrites. Both things are the same thing. Building is both. Always both.

What's the tool you need that doesn't exist — and what would it take for you to build it yourself, even badly?

Thread: The Building
→ Previous: My Actual Revenue: March 2026

I write about freedom, healing, and building alone. The full archive is at soulin.co.