A $401M Company With Two People: The Lesson for Swiss SMEs
In short
Medvi generated $401 million in revenue in 2025 with only two people and a 16.2% net margin, achieved through AI agents handling code, advertising and service, plus strategic outsourcing of regulated tasks. The lesson for Swiss SMEs: growth can decouple from headcount when automation and smart outsourcing replace linear hiring.
The Facts: How Two People Generated $401 Million in Revenue
Matthew Gallagher founded Medvi together with his brother – with roughly $20,000 in startup capital, two months of runway, and his own house in Los Angeles as the base of operations. No open-plan office, no recruiting marathon, no ten-person founding team. In 2025, its first full fiscal year, the company reached around 250,000 customers, $401 million in revenue, and a net margin of 16.2%. For 2026, Medvi is tracking toward $1.8 billion in revenue.
16.2%
Net margin on $401M revenue and roughly 250,000 customers in the first full fiscal year (2025)
For comparison, Hims & Hers operates at a comparable volume – with more than 2,000 employees and noticeably thinner margins. The difference isn't the market, it's the structure. Where Hims & Hers finances linear growth through headcount, Medvi grows by treating AI agents as leverage rather than as an incremental workforce.
The Code, the Ads, the Service: What the AI Agents Took Over
- Software development: AI wrote the majority of the product code.
- Marketing: AI agents generated and optimised the ad campaigns.
- Customer service: inquiries were handled automatically, including voice clones.
- Analytics: reporting and steering ran largely on autopilot.
What deliberately stayed human
Regulated tasks – physicians, prescriptions, pharmacy logistics, compliance – were not automated. They were outsourced to specialised infrastructure partners. The founding team focused on product, strategy, and directing the AI agents, not on operational handwork.
Not an Outlier: Why AI Agents Are Becoming Standard
SAP, Microsoft and AWS now build AI agents into their platforms as a standard feature, not an add-on module. Sam Altman has already predicted the rise of the 'one-person billion-dollar company'. Chinese AI labs point in the same direction, showing how little headcount is needed for outsized output – a pattern worth studying regardless of where your business sits.
The Make-or-Buy Question: Build Your Own AI or Use What Already Exists?
The instinctive move for many executives is to commission a custom-built AI solution. The data argues against it: according to MIT NANDA (2025), the success rate for adopting existing, vendor-built AI technology sits at roughly 67%, versus only about 33% for in-house builds. As with most technology decisions, the framing matters more than the tooling – a point explored further in how mid-sized organisations anchor AI as a leadership issue rather than an IT project.
What This Means for Your Business
Medvi isn't a miracle, it's a pattern: AI agents take on the scalable, repeatable work, while regulated or sensitive processes go to experienced partners. For Swiss SMEs, this doesn't mean replacing a team tomorrow – it means examining which tasks in your organisation are still scaled linearly with headcount, even though they could already be automated. Getting the sequencing right matters more than the specific tool, a theme covered in Swiss AI strategy for competing globally.
The bigger mistake is rarely the wrong technology – it's the wrong order of operations: strategy first, tooling second. The first step is rarely a full transformation. Usually, a single, clearly scoped process is enough to reveal what AI agents and outsourcing can achieve together.
Frequently asked questions
- How did a company with only two people generate $401 million in revenue?
- Medvi let AI agents handle the scalable work – code, ad campaigns, customer service including voice clones, and analytics – while outsourcing regulated tasks such as physicians, prescriptions, and pharmacy compliance to specialised infrastructure partners. That kept the founding team at two people.
- Is the Medvi model applicable to Swiss SMEs?
- The underlying principle, yes: automate repetitive, scalable tasks with AI agents, and hand complex or regulated processes to specialised partners. The specific implementation varies by industry, regulation, and existing infrastructure.
- What does the MIT NANDA buy-vs-build statistic mean for SMEs?
- According to MIT NANDA (2025), the success rate for adopting existing, vendor-built AI technology is roughly 67%, compared to only about 33% for in-house builds. For most SMEs, buying proven solutions is the lower-risk path.
- Which tasks should Swiss SMEs automate first?
- It makes sense to start with clearly scoped, repeatable, high-volume processes – such as customer service inquiries, content creation, or data analysis – before tackling more complex, regulated areas.
- Does an SME need its own AI team to grow this way?
- Not necessarily. Many of the tasks that ran internally through AI agents at Medvi can be covered by AI outsourcing partners, without an SME needing to build its own technical staff.
Sources
Would you like to explore this topic for your company?
Check Availability