AI for Automation
All Business Cases
2026-03-22SaaS AcquisitionMicro SaaSSolo FounderAI BusinessPortfolio StrategyRemote Team

The Man Who Failed at a Delivery Startup Now Makes $120K/Month Buying Other People's SaaS

He co-founded an AI chatbot company and sold it to Microsoft. He joined a startup as employee #9 and watched it grow to $250M ARR. Then he built his own delivery startup — and it failed miserably. 'I am not someone who builds things from zero.' The moment Pascal Levy-Garboua admitted that, he started buying SaaS products built by others. Today, three products generate $120,000 per month.

Pascal Levy-Garboua — Noosa Labs Founder

Pascal Levy-Garboua

@pascallg
23-year tech veteran. Co-founded AI chatbot VirtuOz (→ sold to Microsoft), employee #9 at Checkr ($250M ARR), early investor in Notion. Now acquires and operates micro-SaaS businesses through Noosa Labs while splitting time between San Francisco and Paris.

"I Am Not Someone Who Builds Things From Zero"

A delivery app was quietly dying in a San Francisco apartment.

Pascal Levy-Garboua already had an impressive track record. He co-founded VirtuOz, an AI chatbot company in France, and sold it to Nuance — which was later acquired by Microsoft. He worked at an image recognition startup that Yahoo bought outright to power Flickr's photo search. He joined Checkr, a background-check SaaS, as employee #9 and watched it grow to $250M ARR with 300 employees. He invested early in Notion and earned returns many times over.

Yet the delivery startup he built himself failed. Miserably.

I am not someone who builds things from zero. I am much better at taking something that already exists and bringing it to the next level.

— Pascal Levy-Garboua

That single sentence changed everything. In 2021, he came back with a completely different approach: stop building from scratch. Instead, buy small SaaS companies that others have already built — and grow them.

Five years later, his company Noosa Labs generates $120,000 per month across three micro-SaaS products, with a remote team spread across seven countries, working between Miami and Paris.

$120K
Monthly Revenue (MRR)
$1.44M
Annual Revenue (ARR)
3
Products in Operation
7 Countries
Remote Team Spread

From VirtuOz to Checkr — 23 Years of Silicon Valley Survival

Pascal's journey started in France. After studying at Stanford, he co-founded VirtuOz in the early 2000s — when the word "artificial intelligence" wasn't even in vogue yet. The concept of AI chatbots barely existed in the market at the time.

VirtuOz built virtual agents for enterprises — AI that automated customer service. The technology caught the attention of Nuance, a voice recognition specialist, which acquired VirtuOz. Then in 2022, Microsoft acquired Nuance for $19.7B, embedding the technology Pascal helped create deep into Microsoft's AI infrastructure.

💡 VirtuOz → Nuance → Microsoft

The AI chatbot technology Pascal co-founded passed through two acquisitions and now lives inside Microsoft's enterprise AI stack. He built an AI company nearly 20 years before the AI boom.

He then joined an image recognition startup — which Yahoo acquired to enhance Flickr's photo search. That was his second exit.

Then came Checkr. Joining the background-check SaaS as employee #9, Pascal experienced every stage of the company's journey from startup to $250M ARR. He watched the team grow to 300 people, living through every phase of scaling firsthand.

He also angel invested for 13 years in parallel. His portfolio included Notion and Checkr. He knew how to succeed in Silicon Valley — at least in someone else's company.

The Delivery Startup Failure — Building From Zero Wasn't My Game

So he decided to try it himself. He founded a delivery startup in San Francisco. He'd made VirtuOz work. He'd watched Checkr explode. Surely he could do it again.

It failed.

He hasn't shared the specific numbers, but Pascal says the experience gave him a defining realization: he is not a 0→1 person — someone who creates something from nothing. He is a 1→10 person — someone who discovers what already exists, improves it, and scales it.

🔑 The Pivotal Turning Point

Most startup advice says "build." Pascal chose the opposite. Don't build — buy. The moment a 23-year tech veteran honestly acknowledged his own weakness, a new path opened up.

The Birth of Noosa Labs — A Software Record Label

In 2021, Pascal founded Noosa Labs with a simple, clear mission: buy small, profitable SaaS businesses and run them better.

He calls his model a "software record label." Just as a record label distributes and markets an artist's music, Noosa Labs acquires software built by indie developers and takes over distribution, growth, and operations.

We are not flippers. We improve the product, support the customers, and build on the existing foundation. We want founders to feel confident that when they leave, their customers will be taken care of and their legacy will thrive.

— Noosa Labs official site

Acquisition Criteria — What to Buy, What to Pass On

CriteriaDetails
Annual Revenue (ARR)$200K – $800K
Profit Margin60%+ (excluding founder salary)
Customer BaseSMB or prosumer
Growth MethodPrimarily organic (SEO, word of mouth)
Tech StackStable, maintainable codebase
CulturePositive and ethical team

Pascal is especially emphatic about "platform dependency." He avoids products that rely too heavily on a single large platform's API or policies. Why this matters becomes painfully clear in the story of his first acquisition failure.

The Storm of Year One — 4 Acquisitions, 1 Shutdown, 2 Sold

In 2021, Noosa Labs completed four acquisitions in its first year. Aggressive. But one of them became a disaster.

WAMessages — A Letter From Meta

One acquired product was WAMessages, a Chrome extension that let users send bulk messages via WhatsApp. Revenue looked decent. Users existed.

In December 2022, a letter arrived from Meta (formerly Facebook): a cease-and-desist. The tool violated WhatsApp's Terms of Service.

There was no choice. WAMessages was shut down immediately. A nine-month recovery period followed.

⚠️ The Reality of Platform Risk

Building a business on top of a large platform means handing someone else the kill switch. WAMessages was Pascal's most expensive lesson — and "avoid platform dependency" became a core pillar of Noosa Labs' acquisition criteria from that point forward.

Of the remaining acquisitions, two were sold off. In the end: four bought in year one, one forcibly shut down, two flipped, one kept. A brutal education.

The Surviving Portfolio — 3 Products, $120K MRR

Here's a closer look at the three products Noosa Labs operates today.

1. Sendtric — Email Countdown Timer

The biggest revenue driver in the portfolio. Sendtric lets email marketers embed real-time countdown timers into campaigns. A live "2 hours 13 minutes left in the sale" timer inside an email lifts conversion rates. Simple — and powerful.

ItemSendtric Details
Tech StackRuby on Rails + Go
Growth ChannelSEO + word of mouth among email marketers
Recent ImprovementsEnterprise/API plan added, new widget onboarding
Churn Reduction20–30% annual decrease (new onboarding effect)
Revenue StrategyPrice increases + new enterprise tier

2. Evalart — Hiring Skills Assessment Platform

A platform for running online assessments during hiring — programming tests, aptitude, knowledge, and psychological evaluations. Over 500 pre-built assessments, custom test creation, and webcam-based anti-cheating features.

ItemEvalart Details
Tech StackPHP + React
Revenue Mix1/3 subscriptions + 2/3 pay-as-you-go credits
Growth ChannelGoogle Ads
Supported LanguagesEnglish, Spanish, Portuguese
Growth StrategyMove away from discounts → simplify pricing

Evalart's revenue mix is interesting: only one-third comes from subscriptions, while two-thirds comes from pay-as-you-go credits. Companies run assessments in bulk during hiring seasons, purchasing large credit packages in one shot.

3. Mava — AI-Powered Community Support Platform

The portfolio's highest-growth-potential product. Mava consolidates customer support inquiries from community channels — Discord, Telegram, Slack — into a single unified inbox, with AI automatically answering repetitive questions.

ItemMava Details
Tech StackTypeScript + Svelte
Key Metrics3.5M+ tickets handled, 3,000+ communities
AI Impact60% ticket reduction, 150K+ AI auto-responses
Growth ChannelDiscord bot viral spread
PricingFree – $599/month per workspace
🤖 Revenue Generated by AI

Mava's AI agent learns from existing knowledge bases (websites, Google Docs, GitBook) and automatically answers customer questions. With AI handling 60% of tickets, it cuts support costs in half for customers. Over 3,000 communities are already using it, with more than 3.5 million tickets processed.

Mava's growth engine is distinctive. It starts as a Discord bot — when one community deploys it, other community managers ask "what is that?" and viral spread follows. Outbound sales have recently been added to the mix.

Revenue Growth Timeline

Jan 2021
Noosa Labs founded. Micro-SaaS acquisitions begin. Four deals completed in year one.
Early Stage
Dec 2022
WAMessages forcibly shut down following Meta cease-and-desist. Nine-month recovery period begins.
Revenue Hit
Mid 2023
Portfolio restructured around Sendtric and Evalart. Two products sold off.
Stabilization
2024
Mava acquired. AI-powered community support platform added. Three-product structure established.
Growth Accelerates
2025–2026
Price increases, enterprise tier, onboarding improvements drive 20–30% churn reduction. $120K MRR achieved.
$120K MRR

Tech Stack — Can It Be Maintained After Acquisition?

One of Pascal's top criteria when evaluating acquisitions is the tech stack. He avoids products built on obscure or overly complex technology — because someone needs to maintain it after the deal closes.

Portfolio Tech Stacks

3 products, 3 different stacks — but all "mainstream" technology
Rails + Go
Sendtric
PHP + React
Evalart
TS + Svelte
Mava
7 Countries
Remote Engineers

All three products run different stacks: Rails+Go, PHP+React, TypeScript+Svelte. There's no push to consolidate. Each product keeps the stack it already runs on, and Pascal hires remote developers in seven countries who specialize in each technology. A team spanning four continents operates this way.

🛠 Tech Stack Philosophy

The reason "non-exotic tech stack" is an acquisition criterion is straightforward: you need to be able to hire developers after the deal, and you need to be able to read the code. Rails, PHP, and TypeScript are technologies you can find developers for anywhere.

Cost Structure — Why 60%+ Margins Are Non-Negotiable

The "60%+ margin" requirement in Noosa Labs' acquisition criteria isn't an arbitrary number. After acquisition, there are teams to run, marketing to fund, and products to improve — and there needs to be room for all of it.

Servers / Infrastructure
15%
Payroll (Remote Team)
25%
Marketing (Google Ads, etc.)
10%
Operations / Other
10%
Net Profit
~40%

If roughly 40% of $120K MRR is net profit, that's approximately $48,000 per month flowing through cleanly. This is an estimate — but given Pascal's 60%+ margin requirement at acquisition and the costs layered on afterward, it sits within a reasonable range.

💰 Why $200K–$800K ARR?

In early 2021, acquisition multiples for larger SaaS ($1M+ ARR) were running 4–8x ARR — requiring millions of dollars upfront. Pascal focused on smaller assets where competition was thinner. The $200K–$800K ARR range is too small for large PE funds but the right size for individual acquirers. Less competition meant more reasonable prices.

Marketing Strategy — Invest in Growth Before Product

Focus much more on growth initiatives than on product improvements or new features.

— Pascal Levy-Garboua

This is counterintuitive advice. Most developer-founders obsess over the product — one more feature, another refactor, a UI polish. Pascal says the opposite.

Here's what he prioritizes immediately after an acquisition:

PriorityActionEffect
1stRaise prices and restructure pricingImmediate revenue increase
2ndImprove onboarding process20–30% churn reduction
3rdAdd enterprise tierHigher ARPU
4thOptimize SEO and word of mouthOrganic traffic growth
5thFeature improvements (when necessary)Customer satisfaction

Notice that "feature improvements" ranks last. Getting an already-good product in front of more people at a higher price comes before making the product better.

Each product also runs on a different marketing channel. Sendtric relies on SEO and word of mouth among email marketers. Evalart runs on Google Ads. Mava spreads virally through its Discord bot. Partnerships, Pascal notes, "delivered less than expected."

Finding product-market fit always takes longer than what you read on X (Twitter).

— Pascal Levy-Garboua

Competitive Landscape — The Software Rollup Map

Pascal's model isn't unique to him. Acquiring and operating micro-SaaS has become a recognizable category.

PlayerScaleApproachDifferentiator
Noosa Labs (Pascal)$120K MRR / 3 productsSmall SaaS acquisition + operationsGrowth optimization via founder experience
Constellation Software$6B+ revenueLarge vertical SaaS acquisitionsPortfolio of hundreds of companies
Tiny (Andrew Wilkinson)$100M+ portfolioInternet business acquisitionsStarted as design agency, broad scope
SureSwift Capital$100M+ AUMSaaS acquisition fundInstitutional investor capital
Tibo Maker / John RushIndividual scaleSolo micro-SaaS portfolioUltra-small + automation-first

Pascal credits Constellation Software as his inspiration — the legendary Canadian software rollup that acquired hundreds of vertical SaaS companies to build $6B+ in revenue. He's taken that model and applied it at micro scale.

📊 Acquisition Market Valuations

In early 2021, SaaS with $1M+ ARR traded at 4–8x ARR multiples. Pascal's target range of $200K–$800K ARR trades at roughly 1–3x ARR. A product doing $200K ARR can be bought for $400K–$600K — meaning the acquisition cost is recoverable within two to three years.

Operating Philosophy — Why Not Flip?

The SaaS acquisition market has two types of players: those who buy and grow, and those who buy and flip. Pascal is firmly the former.

His operating philosophy rests on three principles:

💼 Three Operating Principles

1. Customers first: Existing customers should feel no disruption after an acquisition. When the founder leaves, the product and support remain.

2. Incremental improvement: No wholesale rebuilds — instead, gradual optimization of pricing, onboarding, and marketing.

3. Long-term hold: The goal isn't a quick flip — it's compounding growth over five to ten years.

This approach also matters to the founders being acquired. When handing over something you've built over years, "will this person take care of my customers?" is often the deciding factor. Pascal is selling that trust.

Are Vibe-Coded Products Acquisition Targets Too?

Pascal recently shared an interesting perspective on LinkedIn: "vibe-coded" products — built using AI-generated code — are also on his radar for acquisition. His position: product-market fit and revenue matter more than code quality.

⚡ SaaS Acquisitions in the Vibe-Coding Era

In an era where AI writes code, "is the product making money?" may become a more important acquisition criterion than "is the code well-written?" Pascal is already evaluating AI-generated codebases as acquisition targets. This could represent a new exit path for vibe-coding founders.

Marketing Strategy — Invest in Growth Before Product

Pascal's numbers, translated for context:

ItemUSD
Monthly Revenue (MRR)$120,000
Annual Revenue (ARR)$1,440,000
Estimated Monthly Net Profit (~40%)$48,000
Acquisition Target ARR Range$200K – $800K

$120,000 per month in revenue. Roughly $48,000 per month in estimated net profit. All without building a single product from scratch. Of course, 23 years of tech experience, a strong network, and acquisition capital are prerequisites — but the model itself is replicable.

2026 Plans — The Next Acquisition and the Record Label Model

Pascal has outlined three priorities for 2026:

🎯 2026 Roadmap

1. Accelerate growth at Mava and Evalart — deepen AI features, expand enterprise customer base.
2. Scale Sendtric to enterprise — pursue API partnerships with email service providers (ESPs).
3. Complete one new acquisition — expand the portfolio.

More interesting is his long-term vision: evolving the "software record label" model so that indie developers build the product while Noosa Labs handles operations, marketing, and growth. Expansion through partnership, not just acquisition.

Key Takeaways

What Pascal's story offers to any founder or builder:

📝 Lesson 1: Know Your Type

"I am a 1→10 person, not a 0→1 person." That admission was the beginning of everything. The startup world is saturated with "build it" messaging — but acquiring, operating, and optimizing is an equally valid path. For people who are better at growing things than making them, it may actually be the better one.

📝 Lesson 2: Don't Build on Someone Else's Platform

WAMessages was the most expensive lesson. Building a business on top of another company's platform means giving that company a kill switch over your livelihood. Any product that depends heavily on a third-party API — whether Meta, Google, or anyone else — carries this risk.

📝 Lesson 3: Growth Initiatives Before Product Improvements

The biggest trap for developer-founders is product perfectionism — one more feature, one more refactor, one more UI tweak. Meanwhile, the higher-leverage moves — pricing and marketing optimization — go untouched. Pascal generated immediate revenue gains from price increases alone.

📝 Lesson 4: Small Is a Fine Place to Start

A SaaS doing $200K ARR is a small product. Buy it at 2–3x ARR, operate it for five years, and after the acquisition cost is recovered, you have a pure cash flow machine. Acquiring small, automated B2B SaaS products is a strategy that works at any scale.

We are not flippers. We improve the product, support the customers, and build on the existing foundation.

— Noosa Labs

Like Günter Richter, who quit his job in his 50s and built $15K MRR through consulting, or Dmytro Krasun, who dialed back his ambitions and reached $25K MRR, Pascal found his own path. The difference: he declared upfront that he wouldn't build anything — and chose to buy what others had built instead.

"I am not someone who builds things from zero." It took 23 years to admit that. And that admission built $120,000 per month.

Get more AI business stories

We break down how real people are making money with AI

Join Telegram Channel