Last week cracked my brain open — and completely rewired how I build products.
Now I launch and think completely differently.
In 8 days and with 160M tokens, I built yofounder.com solo
Here’s exactly how I did it.
1/20
I’m a product designer. My dev skills? Pretty much just Webflow and a little Bubble.
Now all I need is brainpower, stubbornness, and coffee to launch a full tech product — solo.
Here’s how I pulled it off. And how you can too.
2/20
June 8
I apply to my first-ever hackathon
hackathon.dev by Bolt.
The tagline: “The hackathon for non-devs and vibe coders!”
I join as a skeptic.
I’m convinced AI is just a landing page toy.
My goal? Meme on it, test some prompts, post a hot take. That’s it.
3/20
June 10
I get the builder pack. I write my first prompt — a long-winded spec for YoFounder’s MVP.
I wait 15 minutes. Bolt hums, spins…
I’m stunned.
And then delivers a working version.
4/20
On screen:
✔️ Landing page with senior-level copy
✔️ Sign-up flow
✔️ Dashboard
✔️ Invitation system
To get this from my dev team, I’d need a month of specs + 40 days of full-time engineering.
Here?
One prompt.
15 minutes.
Done.
5/20
First thought: Even if it all goes downhill from here — I just saved 30 days.
What I didn’t know: the rabbit hole had only just opened...
6/20
June 11
I upload 50+ co-founder compatibility questions into Bolt.
UI, autosave, form logic — it’s all generated and running.
I’m floored. It flies.
7/20
June 12
The 66-question assessment is live.
I prompt AI for analysis logic — and it starts spitting out the same "unique" result every time 🤡
8/20
My free 10M tokens run out.
I throw $50 → get 13M more.
Keep fighting AI. It’s stubborn. I’m meaner.
9/20
June 14
Breakthrough.
The AI finally works:
✅ It compares answers
✅ Spots misalignments
✅ Writes partnership insights
Now I just need a QA. And coffee. And sleep.
10/20
I decide to add profile avatars. Small tweak, right?
Nope:
❌ Login dies
❌ Sign-up hangs
❌ Console goes full hellfire
Bolt: "No rollback."
Me: Deep sigh + buying 21M more tokens
11/20
June 15
Bolt asks for a local push to Supabase
First real terminal session of my life. Install Docker, Homebrew, Supabase CLI.
Mac’s crying.
I’m crying.
GPT’s the only one holding it together.
12/20
Next 2 days = SQL warzone.
Migrations fail.
RLS policies clash.
Cycle: supabase db push → edit SQL → push → error Repeat 🔁
13/20
Then Supabase whispers: "You’ve hit your 5GB limit."
Me: 🫠
I pay $25 — and after several more SQL rituals, the dashboard comes back.
Buggy. But alive.
14/20
June 16
Upgrade to 120M tokens.
Bolt finally chills out.
I fix:
• UI logic
• Invite flow
• Metadata
Almost there.
15/20
June 18
I ship the first stable version of http://YoFounder.com
A product I had been dreaming about for a year.
Built in 8 days.
16/20
And this is just the beginning.
I’ve got 40M tokens left until July 10.
Next up:
⏱️User-to-user matchmaking
⏱️Triggered emails
⏱️Monetization
17/20
Bottom line?
You don’t need:
✘ Investors
✘ A CTO
✘ 10 months
You need:
✅ 1 week
✅ $200
✅ A brain full of chaos
✅ Ideally: a QA engineer co-founder
QA engineers are the new gold. Not a joke anymore.
18/20
Now?
These 8 days were the wildest, most productive stretch of my 10-year tech career.
19/20
If you’re building with Bolt — here’s my survival kit:
🧠 Prompt like a spec doc. Ruthless clarity.
🧪 QA = your new best friend
💾 Backup EVERYTHING
📦 Check your Supabase storage
🔁 Caught in a bug loop? Don’t quit. There’s always a way out.
20/20
I’d love feedback from early-stage founders, accelerators, angel groups — anyone helping teams form at day zero.
Try a match. Share how it went.
https://www.yofounder.com/
Interesting how the skillset shifts from traditional coding to system design, prompting. . Curious how you see this scaling as the product grows in complexity. Do you think you’ll eventually need a dedicated engineer, or can AI continue to be your main “cofounder”?
That's great that you documented your whole solo-building journey.
This is very similar to YC-cofounder matching, which I actually used to find my previous cofounder. We worked together for a year but ultimately we decided to go our separate ways. I still think the platform provided immense value and I think there will be more demand for cofounder matching since it definitely is a complex problem.
Founders will spend more time with their cofounders than their actual partners so it is a very specialized matching system.
As a founder, I always struggled with choosing the right co-founder. Platforms like YC Co-Founder Match were helpful for discovery, but lacked a way to validate compatibility.
I wanted something that could help founders talk through differences before they become dealbreakers. That’s where the idea for YoFounder came from.
That makes sense. Yeah YC-Cofounder matching mainly leaves all that stuff to the users once they connect.
Definitely would be interesting to see a platform that outlines specific details of cofounding a business. Actually, YC-Cofounder matching sends you their word doc of how to find a good cofounder that I found very helpful. Would be interesting to have a process where the information in that doc are part of the process of finding a cofounder.
A realistic look at what it actually takes to ship with these tools. Your experience shows it's less AI builds for you and more AI is a new kind of technical partner you have to actively manage and debug. The skill is shifting to system whispering. Great work pushing through it.
Building things as a non-technical person is definitely possible now. But, I think it's mostly from 0 to POC phase. After a while, when the service get's complicated, the LLMs can't keep track. You'd need an engineer for something more serious. But, I guess it's a good start.
The video overview https://x.com/yuliyayugo/status/1935672819276448207