How I Built a $1M a AI Business... Then Rebuilt It
The messy path from a Tiktok idea to a solo SaaS launch, and what I learned about AI, conviction, and chaos.
TLDR… I launched Blotato by using AI to test demand, vibe code my first version, build distribution through content, survive a messy failure-prone launch, then rebuild once I had conviction.
ANNOUNCEMENTS:
(1) to celebrate my birthday this week, get UNLIMITED FREE Nano Banana 2 + 50% off your first month of Blotato with promo code BIRTHDAY2026 (offer valid until 9/1)
(2) I’m pausing my weekly Friday livestream b/c I have lots of family & travel. I’ll be in San Francisco in October!
Most people want the clean founder story:
Idea, build, launch, revenue.
That is NOT how Blotato happened.
Blotato happened because each messy step forced the next one.
A 30-second blurry shaky Tiktok video created demand.
My UGLY AF MVP created proof.
My early adopters created confusion.
My launch mistakes created urgency.
Then the fast initial revenue created enough conviction to rebuild the whole thing.
That is the real story.
Tiktok failed to load.Enable 3rd party cookies or use another browser
The goal was never to vibe code perfect software.
The goal was to create enough evidence to keep going.
1. The Idea Was A Public Question
My August 1 TikTok (above) was the first real test.
I asked if people wanted an AI tool that took YouTube videos, blog posts, and ideas, then turned them into social posts for multiple platforms.
That tiny public question mattered because I wasn’t asking myself if the idea was good.
I was asking OTHER PEOPLE IN THE REAL WORLD about their pain.
The comments told me the pain was real enough to try building it. Very honestly, I figured if Blotato flopped, I could just release the code as open-source and make content out of it! That’s why it felt low-risk and not that serious.
But talk is cheap…
So the next question:
Could I build my small idea FAST, before losing interest?
2. The Vision Was Too Big, So I Cut It Down
On August 4, 2024, I filmed myself building the MVP in a 7-part Youtube series using Cursor AI. This was before Cursor Composer - I literally had to CREATE and DELETE files manually.
Vibe coding tools have come a long way since then.
Y’all have it so good now :)
The first video was not about writing code.
It was about cutting the product scope as much as possible.
My original Google Doc said the vision was:
Social media platforms -> AI-powered REMIX -> social media platforms
It had everything:
Any source: YouTube, Twitter, Facebook, newsletters, articles
Your brain: brand voice, product info, services
AI filtering: quality and relevance
AI remixing: content in your voice
Publishing: any social platform
Weekly queue: Mon/Wed/Fri posting
That vision was exciting.
It was also way too big for v1.
So I drew the first UX prototype with URL input, text input, LinkedIn, Twitter, Facebook, images, edit buttons, and publish buttons.
Then I stripped out the non-essential parts.
No URL input.
No multiple platforms.
No image generation.
No real publishing.
Just text in, Facebook post out.
These are real screenshots from that Youtube series! I knew I had to document my journey, otherwise no one would believe me later.
The Cursor-built MVP looked gross.
But it did the job.
That is the part people misunderstand about vibe coding.
AI did not give me a mindblowing app.
AI gave me a fast first version to start getting feedback.
It gave me momentum.
Then, I invited real users.
And real users immediately made the problem messier :)
3. Early Adopters Proved Interest Is Not The Same As Behavior
On October 1, 2024, I opened an early adopter program with around ~50 ppl.
My original Blotato waitlist form was honest about my stage…
I made it clear: this was NOT a polished product.
It promised the first 50 early adopters a steep lifetime discount and personal, priority support from me, the founder.
It also used my own distribution as a case study.
At that point, I had grown from zero to almost 200k followers in 4 months using AI + automations, while distributing 100+ pieces of content per week across multiple social platforms.
That mattered because Blotato was not an abstract SaaS idea.
It was the productized version of a workflow I was already using.
The form also shows the original positioning problem.
I was building for:
Solopreneurs
Solo creators
Small businesses
Small agencies
Useful audience.
Also broad.
I thought the Google Form answers would tell me what to build next.
I WAS WRONG!
What people said they wanted did NOT match what they actually did.
I realized people wanted a billion different things:
Podcast repurposing
News monitoring
Personal brand posts
Business content
Full automation
Manual editing
Faceless videos
+ 4 million more things
(and now I joke Blotato = BLOATato because I actually tried to cram half these things in there…)
The mistake was treating “interested users” like 1 clean customer group.
They actually had 50 different use cases.
That almost killed my confidence.
Instead, it changed the question again.
Not “is anyone interested?”
Now it was:
Who has enough PAIN to pay, onboard, complain, AND keep using it?
Only a launch would answer that.
Here’s a screenshot of part of my launch email:
4. Launch Day Broke In Public
Official launch day was October 20, 2024.
I sent the email, posted the video, and opened the doors.
Then a bunch of shit broke :D
Tiktok failed to load.Enable 3rd party cookies or use another browser
The worst mistake was the YouTube demo.
I uploaded the wrong video, linked it in the newsletter, sent it to thousands of people, then made the video private!
(I was really nervous okay…)
So the launch email had a giant broken YouTube block at the top.
And that was NOT my only failure, here’s a few more to make you feel better:
Mobile site: broken by last-minute testimonials
Apple Pay: not working
Amazon Pay: not working
Calendar invites: missing meeting links
Honestly???
This is entrepreneurship.
You put the thing out there, and reality immediately punches you in the face.
But my failure-riddled launch still did something important.
It proved people were trying to buy, book calls with me, watch my demo, and sign up for my product.
That is a VERY different signal than a comment.
That is BEHAVIOR.
My barebones launch email stated my core promise:
Blotato was for coaches, consultants, solopreneurs, small businesses, and creators who wanted high-quality content without duct-taping automation tools together.
BTW my website was a $60 Webflow template with swapped text and images. Not some fancy AI-generated 3d-animation mindblowing $10000 website. And I did NOT even update it for 1.5 years… So, don’t be one of those ppl stressing out over a perfect website. Just get on with it!!
Despite my crappy looking website, it was enough because the real learning moved into customer calls, payments, onboarding, and fixing the product.
And that is where my vibe coded MVP hit its limit.
5. Revenue Gave Me Permission To Rebuild
One week after launch, I posted 10 lessons from my first 7 days.
Tiktok failed to load.Enable 3rd party cookies or use another browser
My biggest lesson was launch FASTER.
From August 1 to October 20, I still think I spent too long.
To be fair, I also had multiple birthday celebrations, family visiting, and a road trip to Bryce and Vegas in that stretch of time… but still. I felt I was too slow.
So, don’t do be slow like me :)
The second lesson was technical.
My vibe coded codebase was terrible spaghetti code.
It got me moving, validated demand, and built internal conviction.
Then it became the wrong tool for the next job.
Once the product had real users, I needed a reliable codebase that I understood and could easily maintain and extend. Late 2024 vibe coding tools simply could NOT do this without substantial technical human guidance and constant intervention.
My eventual stack:
Frontend: Next.js
Backend: Fastify
Database: Supabase
Hosting: Heroku
This was the shift from 100% AI coding to AI-assisted coding.
Vibe coding = AI driving and me not looking :D
AI-assisted coding = me driving and AI helping with scaffolding, brainstorming, writing tests, writing comments, coming up with variable names, etc.
Huge difference.
The key point:
I only had the patience & conviction to rebuild because I knew my product made money.
That is the real value of vibe coding a messy MVP.
Not your code.
Your conviction!
From launch day, Blotato has grown linearly in revenue ever since.
HOWEVER:
I usually have a 1.5 month revenue plateau during ski season, late December through January :D
(I like skiing okay smh…)
some months are slower growth than others, especially when I go on vacation a lot (next few months will def hurt)
I don’t feel I have product/market fit like some insane unicorn startup
It still feels like a lot of “pushing” on my side to keep growing revenue
…but damn it’s crazy what you can build with AI!
My business grew steadily and linearly because demand was real, my audience trusted me, and I kept fixing the product in public.
Launching Blotato taught me the order makes a HUGE difference:
Content: test demand in public
AI: build the 1st version fast
Users: find the truth behind what they say
Launch: lots of things will go wrong, get it over with!
Revenue: conviction to rebuild and COMMIT
So if you’re building with AI, don’t worship your v1.
Use it to earn the next piece of evidence.
And if you’re overwhelmed by the process…
The chaos is not a side effect.
The chaos IS the learning :)
Build the ugly proof.
Let real users make everything worse.
Then rebuild the thing that they were willing to pay for, despite all the things that went wrong.
ANNOUNCEMENTS:
(1) to celebrate my birthday this week, get UNLIMITED FREE Nano Banana 2 + 50% off your first month of Blotato with promo code BIRTHDAY2026 (offer valid until 9/1)
(2) I’m pausing my weekly Friday livestream b/c I have lots of family & travel. I’ll be in San Francisco in October!
Not Sure Where To Start? 👇
I’ve organized my best trainings below:
Subscribe to my Youtube and hit the “notification” bell
So that you never miss my free trainings 👇
What was your followers counts on social when you first started building out the mvp
Sabrina, thank you so much for sharing this. I really needed to read it today.
I’m just finishing up my own AI-powered proofreading service, and honestly, I’ve been pretty nervous about putting it out into the world. Reading your story made me feel a lot better about launching before everything feels perfect. :)
Thank you for the encouragement!