- Needs To Exist
- Posts
- He Broke His Spine. Then Shipped an AI Startup From Bed
He Broke His Spine. Then Shipped an AI Startup From Bed
Needs To Exist
August 23, 2025
Some ideas sneak in the side door.
Others crash through the window, demanding attention.
Most die in the quiet gap between “that’d be cool” and “I actually built it.”
This is where It Exists comes in. We chase down the stubborn ones, the sparks that somehow survived the gap. The napkin sketch that didn’t stay a sketch. The half-joke that turned into a login screen.
Sometimes, NTE Pro gave them the raw material—one weird idea buried in a vault of thousands. Sometimes, NTE Zero to One was the push, the map that carried them over that thin, invisible edge between daydream and demo.
Different origins, same destination:
The thought became a thing.
The spark became a story.
The idea? It now exists.
Quentin had the “safe” life down.
Belgian-born, living in Poland. Sixteen years in hospitality IT. Worked his way up to the C-suite.
The kind of guy whose LinkedIn screamed “solid corporate operator.”
But inside? He was restless.
Five years ago, he even flirted with ditching IT to study web dev. But then came the realization:
“I would not enjoy working and fixing bugs on an old codebase.”
So he stayed. Built little projects on the side. Even launched his first SaaS during the GPT-3/DALL·E2 wave, an image generator. It didn’t blow up, but it lit a fuse.
Then came the crash.
An electric unicycle.
Yes, one wheel. No handlebars. Pure midlife-crisis gadget.
The ride didn’t last long. He wiped out. Shattered his spine. Months of rehab.
Most people would’ve drowned in Netflix. Quentin opened his laptop.
Flat on his back, he thought:
“I love building. I love efficiency. I enjoy mapping systems.”
He needed clarity. His idea wasn’t just another side project, it was a tool that solves his own problem.
RedLead is an AI-powered Reddit engagement platform that filters the “needlessly long Reddit rabbit holes” and surfaces only the high-quality conversations worth engaging. Think consultants, coaches, agencies, anyone who wants to catch meaningful signals, not drown in noise.
The 72-Hour Build
He was done wasting time.
By Wednesday it was live.
By the weekend, a stranger paid him real money. The kind of validation most people wait months for, Quentin got in 5 days, flat on his back.
This wasn’t luck. It was the product of habits Quentin had been sharpening for years. He admits:
“I would rather spend 4 hours on a script that fixes a 5-minute problem than do it manually. Because we can repeat and automate that process 10,000 times afterwards.”
That mindset is why RedLead came together so fast and why it didn’t break.
Lesson 1: Backends Don’t Have to Be Scary
Most indie hackers treat backend like it’s calculus class. Quentin leaned straight in.
He didn’t fire up AWS. He didn’t beg a friend to “handle the backend stuff.”
He used n8n, a tool most people treat like an automation sidekick.
Quentin treated it as the engine.
In one night, he had workflows powering a production app. And it wasn’t duct tape, it was reliable.
He says:
“Do not be intimidated by n8n. Just give it a try.”
The unlock? The backend isn’t where you win. It’s just the thing that clears the runway so you can take off.
Lesson 2: The PRD Is the Real Cheat Code
Ask Quentin what really saved him, and he’ll tell you straight:
“No PRD = loop error nightmare and spaghetti code.”
Most builders skip documentation, thinking it’s corporate overhead. Quentin flipped it.
He dictated a messy spec into SuperWhisper. Claude cleaned it. Cursor built it. Suddenly, he wasn’t debugging for days. He was shipping in hours.
He even fed Claude Context7 MCP, a tool that gave it updated docs, so the AI always had the latest context. That one move turned Claude from “hallucinating intern” into a real partner.
He even tested Claude’s limits by cramming frontend + backend into one repo. It kinda worked… until CI/CD turned into a nightmare. Lesson learned: even AI has boundaries.
The faster you write a crystal-clear spec, the faster the AI can write your code.
Skip the PRD, and you’re not saving time, you’re just speed-running your way into spaghetti.
Lesson 3: Community Is Your Co-Founder
RedLead started as a one-off demo for Quentin’s private AI guild. He built a workflow to scrape Reddit into Airtable. Cute experiment. He almost dropped it.
Then he met one of the members in real life. They told him: “This is good, you should take it further.” That was the spark.
When he presented the upgraded version to the group, they leaned in. Asked for features. Pointed out holes. Pushed him forward.
That’s when he realized:
“Your first 10 users are not ‘customers.’ They’re co-founders without equity.”
And when those “co-founders” asked for a new feature? He shipped it in hours. That level of speed builds trust and keeps users hooked.
Without them, RedLead would’ve died in his notes app. With them, it became a product people pay for.
Lesson 4: Copycats Can’t Clone Conviction
Two days after launch, Quentin saw someone buy a near-identical domain and clone his idea.
Old Quentin — the corporate exec — would’ve freaked out. Hired lawyers. Written memos. Worried about “positioning.”
New Quentin just laughed.
“Sure, someone copied it. But they don’t have the conviction. They can’t move like I move.”
Copycats are karaoke singers. They can do your song, but they’ll never have your voice.
Because anyone can steal your landing page.
What they can’t steal is your speed. Your story. Your obsession with users.
Copycats can clone your pixels, but they can’t clone your pace.
What He’s Doing Next
RedLead is live with early users. One messaged him at 1am asking for a feature. By 2am, it was shipped.
He’s not slowing down. Quentin quit corporate, but his old company hired him back as a consultant, proof that even they couldn’t let him go.
Meanwhile, he became an n8n Ambassador for Poland, runs AI Automation Guild, and is already eyeing his next project.
He even jokes:
“I’m looking for my next n8n backend opportunity as we speak. I enjoy the process. It’s a game-changer for prototyping.”
The crash didn’t break him.
It broke him free.
What You Can Learn From Quentin
🛠 Backends don’t have to be heavy. If an automation tool can handle it, let it. The goal is momentum, not infrastructure porn.
📝 PRDs are your superpower. Write a crisp spec, feed it to AI, and watch chaos turn into clean code.
👥 Treat your first users like co-founders. They’ll shape the product before you even know what it should be.
⚡ Conviction > Copycats. Anyone can steal your site. Nobody can steal your belief.
Bonus: What He Thinks Needs to Exist
Quentin believes we’re just scratching the surface with B2B scraping + lead tools. Everyone’s rushing to build scrapers. The real gold rush isn’t the shovel. It’s the refinery. Filtering, enrichment, qualification. That’s where the big winners will be. The scrapers are just the warm-up.
He’s not building it yet. But someone should.
Follow Quentin & His Work
👉 redlead.ai — the Reddit signal tool
👉 aiautomationguild.com — his AI guild
👉 makeautomation.co — his automation business
Here’s the thing: the world doesn’t need another person with a “someday” idea.
It needs builders. The ones who take the rough draft in their head and throw it into the wild.
That’s what makes It Exists worth reading, proof that ideas don’t have to wait.
NTE Pro drops the sparks. NTE Zero to One helps you light the fire.
So maybe the next story isn’t Quentin’s.
Maybe it’s yours.