One day in March 2025, Matthew Gallagher changed one line of code on his website, then headed out for a hike in the mountains near Los Angeles. Midway through his walk, his media agency called in panic: his website had recorded zero orders in the past hour. The line of code he had just deployed had crashed in production, and no one at the entire company could fix it for him. There was no on-call SRE (Site Reliability Engineering) team, no incident response playbook, not even a colleague to restart the server. He had to turn around immediately and rush all the way home to fix the code himself.
That single hour of downtime cost him roughly 200 customers. This slightly chaotic scene is precisely the most authentic side of the “two-person unicorn” myth that has stunned the global tech community in the past 48 hours.
If you haven’t been flooded by the news these past few days, let’s align on the facts quickly: Last week, The New York Times confirmed through audited financial data that Medvi, a telehealth startup, generated $401 million in revenue in its first full fiscal year (2025), amassed 250,000 paid customers, and achieved a net profit margin of 16.2%. Even more strikingly, its annual revenue run rate this year is on track to hit $1.8 billion.
And this unicorn, easily valued above $1 billion, has only two full-time employees from start to finish: founder Matthew Gallagher, and his younger brother Elliot, whom he later hired.
In early 2024, Sam Altman bet Alexis Ohanian, co-founder of Reddit, that AI would spawn the world’s first “one-person, billion-dollar company.” At the time, everyone thought this was a vision at least three to five years away. But reality has compressed that timeline to just nine months.
As a serial entrepreneur and an observer who has long advocated for global opportunities for “super individuals,” I have pored over in-depth reports, court filings, and even FDA warning letters about Medvi over the past few days. I have found that while many social media influencers are wildly sharing this “myth,” they only see the stark contrast between the $20,000 startup capital and $1.8 billion run rate, while ignoring the other side of the coin.
Today, we conduct a deep dive: How exactly did Medvi pull this off? What is its underlying business architecture? And what truly valuable lessons can the deliberately hidden fatal risks teach Chinese global entrepreneurs?
Treat Every Business Function as an API Interface
To understand Medvi’s success, we must first break a misconception: Matthew Gallagher did not “build” a telehealth company from scratch.
What he did was build an extremely efficient customer acquisition and marketing front end using AI, layered on top of two already highly mature healthcare infrastructure providers.
GLP-1 weight-loss drugs (such as semaglutide) are undoubtedly the most explosive consumer healthcare sector of the past decade. It is a perfect market with near-unlimited consumer demand, intense desire from buyers, and severely constrained supply through traditional channels. Wall Street projects that by the end of the decade, the anti-obesity drug market will reach $60 to $100 billion.
In this booming market, following the classic Silicon Valley playbook, Matthew would have raised tens of millions in funding, then built a massive team: hired dozens of engineers to code, recruited hundreds of licensed physicians for online consultations, established a complex pharmacy compliance system, and built a large customer service center.
That is precisely the path taken by Hims & Hers, a public company in the same sector: it posted $2.4 billion in 2025 revenue but employs 2,442 people, with a net profit margin of just 5.5%.
But Matthew chose a completely different route.
He grew up in a trailer park and began reselling katanas on eBay at age 18 — a grassroots entrepreneur with extreme pragmatism and hyper-sensitivity to costs. In his previous startup, he led a team of 60 people and learned firsthand the organizational pain of “more people = higher costs = slower decisions.”
Thus, when launching Medvi, he established a core principle: treat every business function as an API interface or a prompt.
For the asset-heavy, compliance-heavy back end, he directly “rented” infrastructure from two companies: CareValidate and OpenLoop Health. The former provides an “out-of-the-box” telehealth platform and physician network, while the latter handles prescription routing, pharmacy fulfillment, and logistics delivery.
For the asset-light, operation-heavy front end, he treated AI as a full-stack operating system:
Used ChatGPT, Claude, and Grok to build platform code and website copy;
Used Midjourney and Runway to generate massive volumes of ad creative images and videos;
Built customer service systems using ElevenLabs voice technology.
Most crucially, back in 2024, when everyone was still treating AI as a chatbot, he began writing custom AI agents to enable these disparate systems to communicate with each other and automate business monitoring. In the words of a B2B founder on Reddit: “He essentially applied the philosophy of microservice architecture to the organizational design of his entire company.”
Doctors? Call CareValidate’s API. Shipping? Use OpenLoop’s service. Coding? Leverage Claude. When all execution work is API-driven and AI-powered, the founder’s core job boils down to one thing: finding leverage and converting traffic into profit.
A single table illustrates the stunning efficiency gap:
Medvi sells GLP-1 drugs for $179 per month — cheaper than most competitors — with an extremely seamless checkout process and no requirement for in-person clinic visits. Two people generate three times the profit efficiency of a public company, with per capita revenue multiple orders of magnitude higher.
This is indeed a perfect AI-native business model. But so far, we have only heard half the story.
Dancing on a Cliff: The Hidden Systemic Risks
Return to the opening scene of the hiking outage. What happens when the complexity of a system exceeds a single person’s cognitive capacity, and there is no redundant organization to share risk?
Six weeks before The New York Times published its high-profile in-depth profile on April 2, 2026 — on February 20, 2026 — the U.S. Food and Drug Administration (FDA) sent a stern warning letter to Medvi.
The FDA accused the company of deceptive marketing, falsely implying Medvi was a compounding pharmacy for compounded drugs, and using phrases such as “same active ingredient as Wegovy® and Ozempic®” on its website to mislead consumers into believing its products were FDA-approved. Failure to correct the violations could result in seizure or injunction.
More worryingly, this warning letter was absent from all the glowing mainstream media coverage.
Furthermore, due to heavy reliance on AI-generated marketing content, Medvi ran over 5,000 active ads in Meta’s Ad Library. Many ads were exposed for using fake doctor personas (such as a fictional “Professor Albust Dongledore”) and even AI face-swapped “before-and-after weight loss photos.” This directly prompted a bipartisan coalition of 35 state attorneys general to pressure Meta, accusing it of allowing AI-generated fake weight-loss ads to proliferate.
Even in its early days, Medvi’s AI customer service chatbot suffered from severe hallucinations: it quoted artificially low drug prices to customers and even invented entire non-existent product lines. To protect his reputation, Matthew Gallagher had no choice but to honor these fake prices.
This reveals the weakest link of the “one-person company” model: when a company of just two people serves 250,000 medical customers, the founder becomes the only human backstop for all system failures.
Moreover, this risk is rippling upstream to the supply chain. OpenLoop Health, Medvi’s heavily relied-upon infrastructure provider, suffered a major cybersecurity incident in January of this year, with data from approximately 1.6 million patients stolen. Later, in a class-action lawsuit filed in North Carolina, plaintiffs accused OpenLoop and affiliated pharmacies of selling oral tirzepatide tablets as “modern snake oil,” since this macromolecular peptide drug is broken down by digestive enzymes before reaching the bloodstream and cannot be effectively absorbed. In this lawsuit, Medvi was named as just one of dozens of consumer front ends operated by OpenLoop.
This is the full picture of the “one-person unicorn”: extreme efficiency paired with extreme fragility.
From B2B to A2A: The Dimension-Reduction Edge for Chinese Entrepreneurs
For all its controversy, Medvi represents an irreversible business trend. For Chinese global entrepreneurs, the lessons of this model far outweigh its flaws.
Over the past few months, a quiet policy race over OPC (One Person Company) has broken out across China. Hangzhou has even earmarked a special annual fund of 100 million yuan, aiming to attract 1,000 OPC entrepreneurs by 2026.
In a 30-square-meter apartment in Shenzhen, Lin Yuan, a former operations manager at a foreign trade company, is replicating Medvi’s success with AI — only in cross-border e-commerce. He uses ChatGPT to write native Russian product descriptions, Midjourney to design main product images, and even AI digital humans for Russian voiceover videos. Most impressively, he had AI crawl 5,000 negative reviews for “warmth” products on Russian e-commerce sites over three months, pinpointing two core pain points: “finger dexterity” and “battery life,” then directly connected with domestic factories to improve the products.
Last winter alone, he sold 7,000 pairs of heated gloves — all by himself.
On TikTok Shop, another Chinese seller PD has automated the entire short-video commerce workflow with AI: agents automatically scrape trending topics, analyze competitor strategies, reverse-engineer prompts, call video generation tools to create content, and finally launch ads automatically.
Kuo Zhang of Alibaba summarized this phenomenon in a recent article as the “collapse of the Execution Wall.” In the past, what held back solo entrepreneurs from scaling was the tedious but critical execution details: from VAT compliance in Marseille to logistics procurement in Shenzhen. To cross this wall, you had to raise funding, hire people, and build massive back-office departments.
But today, with the maturation of Agentic AI, business collaboration is evolving from B2B (Business-to-Business) to A2A (Agent-to-Agent). A buyer’s AI can communicate directly with a seller’s AI via APIs, completing cross-border negotiations and logistics coordination in minutes that once took weeks of human work.
When “execution cost” approaches zero, a single entrepreneur can instantly gain operational reach comparable to Fortune 500 companies. This is not a metaphor — it is a structural reality unfolding right now.
Three Brutal Truths for OPC (One Person Company) Founders
As entrepreneurs, we love the inspirational stories of “one person taking down an industry.” But after deeply analyzing Medvi and China’s OPC ecosystem, I want to share three brutal truths:
First, do not romanticize the “one-person company” narrative — see the structure behind it.
Seventy percent of Medvi’s success stems from GLP-1, a once-in-a-decade perfect market; 20% comes from its brilliantly designed business model (renting infrastructure); and only 10% from its AI toolstack.
Yet that 10% of AI leverage is what made the other 90% possible. AI is not magic — it is simply an extremely cheap computing lever to amplify your market insight. If you pick the wrong product or your business model is unviable, no amount of advanced AI agents can save you.
Second, the core competencies of global entrepreneurship are being redefined.
In the past, the barrier to going global was “resource integration”: you needed a team, capital, and overseas channels. Individuals often exhausted themselves on internal friction from team-building and structure-setting.
But in the A2A era, the entrepreneurial barrier has shifted from “resource ownership” to “depth of cognition.” As Kuo Zhang notes, future competitive advantage will no longer come from technical proficiency or large payrolls, but from judgment, taste, and strategic vision. AI can execute workflows, but humans must provide direction and quality control.
Your bottleneck is no longer insufficient production capacity — it is a lack of imagination.
Third, Chinese entrepreneurs have unique advantages, but compliance is a matter of life and death.
China has the world’s most complete supply chain, and our entrepreneurs rank among the global leaders in AI tool adoption and application innovation. Combining these two strengths, China is highly likely to produce a cohort of truly influential “one-person unicorns” in global markets.
But Medvi’s lessons are clear. When your business scales explosively via AI leverage, your compliance, legal, and ethical risks multiply exponentially. In heavily regulated sectors such as healthcare and finance, using AI to generate fake doctor endorsements or false efficacy comparisons is tantamount to drinking poison to quench thirst.
Being asset-light and operation-intensive does not mean being unprincipled. As AI grants us enormous power, building matching “human backstops” and risk firewalls is a question every OPC entrepreneur must answer.
Returning to the opening scene. Matthew Gallagher, who rushed home from the mountains, eventually admitted one truth: he is considering hiring more people, because extreme efficiency often comes with extreme loneliness.
The “one-person company” is never the end goal of entrepreneurship — it is merely a brand-new starting point given to us by the AI era. For the first time, we have the chance to test a great business intuition without compromising to capital or bureaucracy.
The real question is no longer “Can I do this alone?”
It is: How far can the boundaries of my cognition reach in the world?
No comments yet.