- Startup Spells 🪄
- Posts
- Rosie AI Playbook: How to Source a $1M ARR Idea from YC Startup Directory
Rosie AI Playbook: How to Source a $1M ARR Idea from YC Startup Directory
PLUS: How SaaS companies can sell "unused capacity"
Startup Spells 🪄
November 22, 2025
Rosie AI Playbook: How to Source a $1M ARR Idea from YC Startup Directory
Rosie AI hit $1 million in annual recurring revenue just 8 months after launch by refusing to wait for a lightbulb moment.
Most founders believe you need inspiration to build a massive company. Jordan Gal at Rosie AI proved otherwise.
Rosie AI - Voice Answering Service
After burning $18 million on Rally, he stopped trying to be a visionary. He switched to a ruthless, repeatable sourcing strategy that took him from failed pivot to $1M ARR with only 6 people.
Here's the full autopsy of the $18M mistake and the VC Thief cheat code he used to build Rosie AI.
Rally's Death: An $18M Lesson in Mimetic Failure
Jordan built CartHook to $6M ARR before Shopify crushed it. For his next act, he raised $18 million for Rally, a headless checkout for e-commerce.
20 employees. Millions in the bank. And he hated every minute.
"It felt like everything was harder than it should have been."
The Mimetic Trap: Copying Success That Wasn't His
The diagnosis: mimetic failure.
"Many of my mistakes were mimetic in nature. I was trying to do something that was successful for somebody else."
He copied the standard Silicon Valley Playbook. Enterprise SaaS. Long sales cycles. SLAs. Security questionnaires. Annual contracts. The stuff serious founders do.
But he ignored his own DNA.
When Product-Market Fit Doesn't Equal Founder-Market Fit
Rally forced him into operational hell:
Enterprise sales cycles that dragged for months
E-commerce culture felt like a young guy's hustle game
He didn't identify with the people he was selling to
He had Product-Market Fit. He lacked Founder-Market Fit.
How Jordan Gal Cut His Team by 70% (And Moved Faster)
With a few million left in the bank, Jordan made the call. He gathered his 20-person team for a virtual meeting:
"Do you want to keep doing this? If you're in, cool. If you're not, totally understandable."
6 people stayed. 14 left.
He stopped caring about vanity metrics on TechCrunch. Amount raised. Partnerships.
He focused on his After-Tax North Star: money in his personal bank account, not the company slide deck.
The YC Startup Directory Signal: How He Stole the Winning Idea
Jordan sat there with 6 people and money in the bank, but no idea.
He didn't have a missed call problem. He was a tech founder, not a plumber. He couldn't scratch his own itch.
Instead, he used the YC Signal.
Stealing R&D from Smarter Investors
Jordan went to the YC Startup Directory. He scrolled through recent batches looking for patterns.
Smart money was pouring into AI Voice.
YC Startup Directory - AI Voice
But the products had 2 flaws:
Built for developers or enterprise call centers
They sounded terrible
Bad Tech (Voice AI) Today Means Market Dominance Tomorrow
Most founders see bad technology and walk away. Jordan saw a green light.
"I figured any individual category in AI that was currently bad, was inevitably going to be very good in the near future."
If he built the infrastructure now while the tech was imperfect, he'd own the market position the moment LLMs and voice models upgraded.
He didn't need to invent the tech. He needed to package it first.
How Rosie AI Wins Without Being the Best Technology
He had the category. He needed customers. He refused to go back to e-commerce hustlers.
The Two-Filter System for Validating Ideas Fast
Jordan grew up in an immigrant household running a property tax reduction business. He felt comfortable with family businesses. SMBs. Local services. Home service providers.
He spotted massive Ecosystem Tension:
Tribe A (AI Innovators): Obsessed with tech, ignoring the boring real world
Tribe B (SMBs): Plumbers, painters, martial arts dojos. They have phone problems but can't use complex APIs
The Saturday Morning Avatar: Selling Relief, Not Technology
He defined the product not by technology, but by a specific moment of anxiety:
"It's 10:00 a.m. on Saturday. You're with your family having breakfast. The phone rings."
If the owner answers, they're a bad parent. If they don't, they might lose a $10,000 job.
The value wasn't AI. The value was resolving that tension.
Going Horizontal: The Mailchimp of Voice Strategy
While competitors niched down (AI for Dental Offices), Jordan went horizontal. He built Rosie AI to be the Mailchimp of Voice: accessible, low-cost entry point for any small business.
The Tech Stack:
Stuck to Laravel (PHP) with his longtime CTO
Used Bland.ai for telephony and voice intelligence
Operated as a wrapper so his 6-person team could focus purely on UX
The Market Validation:
1,000 cold emails per day for 2 months
Used Clay and Mailreef
Proved market desperation before writing scalable code
By stealing the category from YC, stealing the R&D from infrastructure providers, and applying it to the boring SMB market, Jordan Gal turned a failed fintech company into an AI rocket ship.
Rosie AI hit $1M ARR in 8 months by looking for patterns in YC Startup Directory.
Top Tweets of the day
1/
I do this all the time. My trick is to rewrite sentences (or sometimes paragraphs) that don't make sense or that I can't articulate properly and using AI to write 5 variations of it.
2/
Many creators depend only on sponsorships. But audiences don't trust you enough and you make less money if you sell a new thing every single time. Often such creators are great opportunity for partnerships. Either cash payments or percentage of revenue.
3/
Talk so you can get more work done. I'm sure we are not far away from having one person cubicles in offices so that everyone can talk rather than type.
Typing now feels so slow even though I'm probably in the top 0.001% of typists out there.
Rabbit Holes
What’d ya think of today’s newsletter? Hit ‘reply’ and let me know.
Do me a favor and share it in your company's Slack #marketing channel.
First time? Subscribe.
Follow me on X.
Reply