Romàn made $7,000 before he wrote a single line of code.
He cloned a Shopify app, grew it to $500,000 a year, and sold it for seven figures only two years later. Then he did the whole thing again, but flipped the order: sell first, build later.
And here's the kicker. He's not a developer. He's a mechanical engineer who got bored, jumped from business to business, and finally decided to "settle" into SaaS.
His second app, GojiBerry, went from $0 to $50K a month in 6 months. Almost entirely through cold outreach. No paid ads at the start. Just relentless distribution.
I sat down with Roman on the Profitable Founder Podcast. Here's the full 90-day playbook.
The $0 mistake that cost him 6 months
Roman's first SaaS was a disaster. And he'll tell you exactly why.
He found a US company called One Text that had just gotten into Y Combinator. Text-to-buy by SMS. At the time he thought YC meant guaranteed success. (He now knows most of those companies fail.)
So he tried to copy it for Europe. SMS buying wasn't a thing here, so he switched it to WhatsApp and spent months rebuilding everything.
They launched. One or two pilots signed up. Both quit fast. European customers were not ready to pay by saying "yes" in a text.
→ 6 months of building
→ Zero real customers
→ One lesson that changed everything
"If we had talked to customers for one week, we would have known not to launch that," he told me. "Everything starts with the customer, not with the product."
Clone what already prints money
The fix was simple. Stop guessing. Find companies already selling WhatsApp tools in Europe and making real money. Then copy them.
He found a Shopify app doing $2-3 million a year. (One founder was literally posting his revenue on LinkedIn. For the rest, Roman used a Chrome extension called Store Leads to see who was installing which apps.)
Here's the move:
→ Search "WhatsApp" in the Shopify app store
→ Create an account on every competitor
→ Pick the one easiest to clone with what they'd already built
→ Ship two main features in 2 weeks
Then he re-contacted everyone from the failed product. Almost no one was selling this in France. Within two days, five to ten clients wanted to pay.
That became Coco AI. This is exactly the kind of copy-a-proven-playbook approach to building a micro-SaaS that beats inventing something from scratch.
The cold outreach machine: 70% close on demos
Roman is a marketer, not a coder. His co-founder handles the product. So distribution was always the weapon.
His daily list while growing Coco AI looked insane:
→ 100 calls per day
→ 1,000 to 2,000 cold emails
→ LinkedIn outreach on top
→ Demos
The whole game was just getting demos. Because once he got someone on a demo, he closed 60 to 70% of them. Get the demo, win the deal.
His ICP was dead simple: a Shopify store doing more than 10K per month in Europe.
And his best hack? He stopped emailing the founder. Instead he emailed customer service with a demo video and asked them to forward it to the CEO. They almost always did.
Bonus: customer service replies to nearly everything, which kept his domains out of spam and let him send serious volume.
Coco AI hit $500K ARR. Then someone slid into his LinkedIn looking to buy a business in that exact revenue range. Two months later, he sold for seven figures. (He took multiple hundred thousand for himself, 90% upfront, 10% over three years.)
Sell first, build later: $7K before a line of code
After the exit, Roman noticed what actually made Coco AI work: reaching out to high-intent leads. People commenting on WhatsApp and email marketing posts replied way more.
So for his next app, he built the offer before the product.
He sent around 20,000 emails saying "we can get you high-intent leads, reply if interested, refund if it doesn't work." Calls poured in. He charged $200-300 a month for 100-200 leads.
The first version was fully manual. He found the leads himself. Then his co-worker helped. Then he hired people on Fiverr with an exact blueprint.
→ Dental clinic hiring a nurse? Type "dental clinic nurse" on LinkedIn
→ Pull everyone interacting
→ Extract the right ones
→ Send the list
He hit $7K MRR before building anything. Then it got unmanageable, he shut it off, built the automated platform, and turned it back on. That became GojiBerry.
"Having money before you code is a great sign," he said. "It means you're really onto something."
Relentless distribution: 11 million Reddit impressions in 4 months
GojiBerry V3 went from $0 to $33K a month in 4 months. No paid ads. Just stacking channels.
Roman ignores the "focus on one channel" advice. His process: post a lot on one channel, learn how it works, then add another. And another.
Reddit was the engine early on. 11 million impressions in 4 months. Tens of thousands of visitors. Free.
The Reddit formula:
→ Always tell a story
→ Back it with proof (screenshots, real numbers)
→ Always include a link
One post that crushed: they applied to Y Combinator, got rejected after the interview, and wrote up the whole thing. What they did, what YC said, the rejection proof, and a link. People ate it up.
His other channels, ranked by impact:
→ Cold outreach (the #1, because you know in days if it works)
→ LinkedIn influencers (cheap, one viral post pays for itself)
→ Reddit, X, YouTube, SEO
→ TrustMRR ads (50-200 visitors a day, plus retweets from big founders)
→ Newsletters (profitable but slow, and hard to track)
→ Affiliates (30% lifetime, climbing to 40-60%)
One outreach tweak worth stealing: instead of asking for a demo, he offers a free 10-page "blueprint" packed with value. Saying yes to a blueprint is easy. Most people never book a call. They just go try the free trial.
Today he sends about 5,500 messages a day across five LinkedIn accounts and Instantly AI. GojiBerry has around 500 customers.
I interview founders like this every week → Watch the Podcast
His exact 90-day plan if he started over
I asked Roman: I want to build a B2B SaaS today. What do I do for the next 90 days?
→ Figure out your skills and unfair advantage
→ Study the market: are competitors raising or bootstrapping?
→ Open PowerPoint, write 7 slides (logo, pain, solution, how it works, price)
→ Message 100-500 people, get them on calls, ask about their problems
→ Get someone to pay. Deliver it as a service first if you can't build yet
→ Validate with 5-10 paying people, then decide if you go all in
Then the marketing rhythm:
→ Month 1: cold outreach only ($300-500 in tools)
→ Month 2: outreach + inbound (post on LinkedIn, X, Reddit)
→ Month 3: add paid traffic and influencers
"When sending 10 messages gets you one demo, send 100. Then 1,000," he said. The 20% that drives 80% is just doing more of what works.
Frequently Asked Questions
Who is Roman and what does he build?
Roman is a French mechanical engineer turned SaaS founder. He started in 2023, built Coco AI (a WhatsApp marketing tool for Shopify stores) to $500K ARR and sold it for seven figures. He now runs GojiBerry, a go-to-market and cold outreach tool with around 500 customers.
How much did Roman make selling Coco AI?
He sold Coco AI for an undisclosed seven-figure sum after growing it to $500K ARR in roughly two years. He personally took multiple hundred thousand dollars, with 90% paid upfront and 10% spread over three years. He says it left him able to not work for years if he wanted.
How did GojiBerry grow so fast?
GojiBerry hit $50K MRR in 6 months, mostly through cold outreach (5,500 messages a day) plus stacked channels like Reddit (11M impressions in 4 months), LinkedIn influencers, X, and TrustMRR ads. Roman validated the idea by selling lead-gen as a manual service first, hitting $7K MRR before building.
The pattern is the same every time: find the customer first, sell before you build, then drown the channel that works in volume.
