A Two-Person Company Is Now Doing $3M a Day — And the FDA Has Questions

Sam Altman has been predicting for two years that AI would produce the first one-person, billion-dollar company. He said it in interviews, on podcasts, to anyone who'd listen. It was the kind of prediction that sounded cool at parties and slightly ridiculous on reflection.

Last week, the New York Times profiled Medvi, a GLP-1 telehealth company built by Matthew Gallagher with $20,000 in startup capital, and the prediction effectively came true. $401 million in 2025 sales. Projected $1.8 billion in 2026. Two employees — Matthew and his brother Elliot. Roughly $3 million a day in revenue right now. Grown from 300 customers to over 250,000 in under two years. A 16.2% net margin.

The AI stack that did it is the kind of thing every solo operator on this site already uses. ChatGPT, Claude, and Grok for code. Midjourney and Runway for ad creative. ElevenLabs for voice. A pharmacy network and prescribing doctors outsourced through CareValidate and OpenLoop. Gallagher didn't write a line of code he couldn't have asked an AI to write. He didn't run a single ad he couldn't generate in minutes.

This is the story every techno-optimist blog is covering as "the solo billionaire era has arrived." And they're not wrong that the business model is real, replicable, and worth studying carefully. But there's a second story running in parallel that didn't make the NYT's headline, and that story is arguably more important for solo operators trying to learn from this. Before the profile ran, the FDA had already sent Medvi a warning letter.

The Playbook That Actually Worked

Let's start with what's genuinely new here, because the noise about this story has been loud enough to drown out the signal.

Medvi is not "one guy wrote an app and became a billionaire." The code is a small part of it. What Gallagher actually did — and what's genuinely new as a playbook — is orchestrate a commercially-viable business using AI as the labor input for every function that used to require hiring.

Marketing copy: written by AI. Ads: generated end-to-end by AI. Voice agents for customer support: ElevenLabs plus a prompt stack. Website and funnel pages: stood up in days. Back-end code and operational tooling: built with Claude and GPT. Everything else — prescribing, pharmacy, shipping, regulatory paperwork — outsourced to specialized B2B vendors.

The insight is that the solo operator is not the person who personally builds everything. The solo operator is the person who assembles a network of AI tools, outsourced services, and automated workflows that together do the work of a twenty-person company. That's the pattern. It's real. It works.

If you strip out the specific industry, the playbook looks like this: pick a domain where regulatory compliance can be outsourced to specialist B2B vendors, build the customer-acquisition funnel entirely with AI-generated creative, automate the customer-service layer with voice and chat agents, and operate as a thin orchestration layer between the customer and the outsourced specialists. This is a repeatable blueprint. A dozen industries could support a dozen Medvis.

That part is worth taking seriously, and it's why Gallagher's story matters beyond the specific company. It's the first large-scale proof that a two-person orchestration layer can capture the kind of revenue that used to require a real company.

The Part the NYT Profile Underplayed

Before the NYT published the profile, the FDA had already issued Medvi a warning letter. That isn't a minor footnote. FDA warning letters are the formal first step in a regulatory enforcement action. They're reserved for conduct the agency considers materially non-compliant, and they typically give the company a set window to respond before further action is taken.

The NYT mentioned the warning in passing. Most of the downstream coverage — the techno-optimist takes, the "one-person billion-dollar company is here" celebrations — either omitted it entirely or buried it below the fold.

The honest read is that Medvi's growth is not separable from its regulatory exposure. The same speed that lets a two-person company ship $3M a day in prescriptions is the speed that skips or compresses steps that exist for reasons. You can't orchestrate "proper patient intake" the same way you can orchestrate ad generation. Some parts of healthcare work the slow, careful way because the fast, careless way kills people. The FDA's concerns aren't paperwork; they're about whether the prescribing process Medvi was using actually constituted medicine as the law defines it.

This is what happens when you apply the one-person-company playbook to an industry that has regulators. Eventually, the regulators notice.

None of this makes Gallagher a villain. The incentive structure in AI-first startup-land is precisely designed to produce exactly this behavior. When the entire venture ecosystem is chanting "move fast, AI has changed the calculus, first-mover advantage is everything," you get companies that move fast, skip steps, and hope the regulatory catch-up is slower than the revenue ramp. Medvi is the cleanest case of that logic playing out in public.

The question for the rest of us is what lesson we actually extract from it.

What Solo Operators Should Copy and Not Copy

The parts of the Medvi playbook that generalize cleanly and are worth copying:

The orchestration model. Build yourself as a thin, AI-powered layer that plugs into specialized B2B vendors for the parts of the stack that are hard to do well. This is genuinely powerful and not specific to telehealth.

The AI-generated marketing flywheel. Copy, images, video, voice — all of it can now be produced at a quality that converts. If your product-market fit is real, AI-generated creative will let you test hypotheses at a pace traditional agencies can't match.

The automated customer-service layer. Voice agents and chat agents are genuinely good enough in 2026 to handle 80% of recurring inquiries for most consumer products. That's a real cost structure change.

The $20K launch budget. Gallagher didn't need capital to get started. The tools are cheap enough that you can test whether your business has legs for less than a used car.

The parts that are not worth copying, no matter how good the numbers look:

Regulatory blindness. If your industry has regulators, the "move fast, skip compliance" approach ends the same way every time. You might get the $401M year first. The letter comes eventually. Planning your business around the assumption that it won't is a category error, not a strategy.

The pretense that orchestration equals expertise. Gallagher doesn't know medicine. His outsourced doctors and pharmacies do. That gap is fine when the orchestration is truly at arm's length; it becomes a liability when the outsourced partners trust you to handle things you've decided to handle with an AI prompt instead of a subject-matter expert.

The "fastest-growing company in history" framing. The moment anyone, including you, starts describing your company in world-historic terms, you've lost the plot. Medvi got profiled in the NYT partly because of the numbers and partly because Gallagher had the skill to tell a very compelling story about them. Those are two different things. One scales with substance and one scales with narrative, and the second one has a much shorter half-life.

The Real Takeaway

The one-person billion-dollar company is a real thing now. That much is true, and it's worth internalizing. The tooling has reached a level where the orchestration-layer-plus-AI pattern can produce outcomes that used to require a hundred people. If you're building solo, the ceiling on what you can plausibly ship is higher than it's ever been, and 2026 is the year that stopped being a prediction and started being documented.

But the first company to break that ceiling did it in a regulated industry, using the speed advantage to outrun the regulators, and it's currently in the FDA's crosshairs. That's not a coincidence. That's the shape of the opportunity. The fastest way to get there is almost always the most compromised way, and you should take the cautionary version of this story more seriously than the celebratory one.

The right lesson is the boring one: the playbook works, the tooling is real, the opportunity is genuine — and the version of this story worth building is the one where you're still in business in five years because you picked a domain where you could move fast without leaving bodies in the road.

Sources

• How one man used AI to build a fast-growing company selling GLP-1 — NewsNation
• Techmeme: Medvi hits $401M in 2025, tracking $1.8B in 2026 (Erin Griffith / NYT)
• MEDVi FDA Warning Letter and $1.8 Billion NYT Profile — HealthDataConsortium
• The New York Times spotlighted MEDVi. The FDA had already warned the self-proclaimed 'fastest growing company in history.' — Drug Discovery Trends
• The 1-Employee Billion-Dollar Startup: How AI Is Changing the Face of Solopreneurship — Inc.