In September 2024, Matthew Gallagher sat down at his desk in Los Angeles with $20,000, no engineering team, and an idea. Eighteen months later, his company — Medvi, a telehealth platform for GLP-1 weight-loss drugs — is on track to generate $1.8 billion in revenue. His total headcount: two people. Him, and his brother.
This isn't a story about a lucky niche or a viral moment. It's a story about what becomes possible when someone with domain knowledge learns to direct a build — using AI as the workforce.
And it comes with a warning.
The Build: How Gallagher Turned Domain Knowledge Into a Platform
Gallagher is not a software engineer. He is not a developer. By most traditional startup standards, he had no business building a healthcare platform.
But he understood the business — telehealth, GLP-1 medications, patient acquisition, compliance flows — and he used that knowledge to direct AI tools to build what he needed, layer by layer.
"It's not an AI company, but I did it with AI." — Matthew Gallagher, New York Times
Step 1: Build the Platform First
Gallagher used large language models — ChatGPT, Claude, and Grok — to write the majority of Medvi's codebase. With no engineering payroll, he iterated on the platform's UX, back-end logic, and telehealth compliance flows through prompts. When something broke, he fixed it. When a feature was missing, he built it.
He didn't need to know how to code. He needed to know what the platform had to do — and he did, because it was his industry.
Step 2: Automate the Operational Layer
Marketing copy, email sequences, ad creative, and SEO content were all generated and tested with AI. Customer service — traditionally the bottleneck for any healthcare business at scale — was handled through AI-powered chat and automated response systems. With over 500,000 patients served, this would have been impossible without modern LLMs.
Step 3: Outsource What You Can't Own
Gallagher didn't try to build a pharmacy or hire doctors. He rented the most expensive, regulated parts of the healthcare stack — clinical infrastructure, prescriptions, and shipping — through platforms like CareValidate and OpenLoop. That freed him to focus entirely on the front end: the platform, the brand, the customer experience.
The result? In its first full year, Medvi generated $401 million in sales at a 16.2% net margin. For context, Hims & Hers — a well-funded competitor with over 2,400 employees — posted a 5.5% net margin the same year.
The Cautionary Tale: When Ambition Outruns Ethics
Here's where the story gets complicated.
As Medvi scaled, scrutiny followed. Reports emerged of AI-generated doctor profiles — photos and credentials that didn't match real individuals — used in marketing materials. Multiple medical providers filed lawsuits claiming their likenesses were used without permission. AI-generated before-and-after images and fabricated ad content have been cited as misleading at best, fraudulent at worst.
The technology Gallagher used to build Medvi's platform was legitimate and brilliant. The shortcuts he allegedly took in marketing were not — and they're now threatening everything he built.
This is the trap that catches builders who move too fast and stop asking "is this right?" alongside "does this work?"
Gallagher proved that a solo founder with domain expertise and AI tools can build a world-class platform. That achievement is real and worth studying. But he may have also proved that scaling at this speed, without accountability guardrails, can turn a compelling origin story into a cautionary one.
The platform itself was the innovation. The fraud allegations have nothing to do with AI's capability and everything to do with the choice to cut ethical corners under growth pressure. Those are two very different things — and they shouldn't be confused.
What This Means for You: The Build to Fit Framework
The Medvi story is proof that the gap between "I have an idea" and "I have a running platform" has collapsed. You no longer need a technical co-founder, a six-figure development budget, or months of runway before you have something real.
What you need is a framework. This is what we call Build to Fit — the idea that your domain knowledge is the blueprint, and AI is the construction crew.
Principle 1: Start With the Problem, Not the Product
Gallagher didn't start by asking "what can I build with AI?" He started with a clear business problem: people need access to GLP-1 medications, and the existing process is slow and expensive. The platform was the solution — not the starting point.
Before you open a single AI tool, be brutally clear on: Who is the platform for? What one workflow does it solve? What does success look like in week one?
Principle 2: Your Domain Knowledge Is the Moat
Ironically, one of the criticisms levelled at Medvi is that it has no proprietary moat — no exclusive technology, no unique supplier relationships. And that's true. But Gallagher's real moat, at least in the early stages, was that he understood the telehealth industry well enough to know exactly what to build.
That's the Build to Fit edge. You are not competing on code. You are competing on knowing your business better than any developer could. AI writes the code. You write the requirements — and requirements built on deep domain expertise are hard to replicate.
Principle 3: Build the Minimum Valuable Platform First
Gallagher launched Medvi in two months. He didn't build everything — he built the one thing that mattered: a platform that could connect patients to prescriptions. Everything else came later.
Your first build should answer one question for one user. Not five questions. Not a full product suite. One thing, working, live.
Principle 4: Automate the Repeatable, Humanise the Critical
Where Gallagher succeeded operationally was in knowing which parts of the business to automate and which to protect. AI handled customer service at volume. Clinical compliance was outsourced to specialists. Marketing was AI-generated and tested fast.
Where the wheels came off was when automation crept into areas that required human judgment and ethical accountability — like who you are and what you're selling.
The rule: automate the repeatable. Never automate the trust.
Principle 5: You Don't Need to Know How to Code. You Need to Know How to Direct.
This is the core of Build to Fit. Gallagher didn't write a single line of code himself. He wrote prompts. He reviewed outputs. He tested, iterated, and made decisions. That's what directing a build looks like in 2025.
The skill is not technical. It is managerial — knowing what to ask for, how to evaluate what you receive, and when to push back. Those are skills you already have. You've been using them in your business for years. AI just made them applicable to software.
Where to Start This Week
You don't need $20,000. You don't need a brother. Here's a concrete starting point:
- Write down the one internal process in your business that costs you the most time or creates the most errors.
- Describe that process in plain language — who does it, what triggers it, what the output should be, and where it breaks.
- Open an AI tool and paste that description. Ask it to suggest a simple tool or workflow that would automate or improve it.
- Don't ask it to build everything. Ask it to build step one.
- Ship step one. See what breaks. Build step two.
That's it. That's the beginning of a platform. Gallagher went from step one to half a billion dollars. Your ceiling is your own.
The Takeaway
Matthew Gallagher built something remarkable. A custom telehealth platform, from scratch, with no engineering team, generating over half a billion dollars in verified revenue. That is not luck — that is what happens when deep domain knowledge meets the right tools and relentless execution.
The fraud allegations are a separate conversation — one about ethics, accountability, and the dangers of optimising for growth above all else. You can admire the build and reject the behaviour. Both things are true.
What matters for you is the lesson underneath: the barrier to building is no longer technical. It is clarity. Know your problem. Know your user. Know what you're building and why. Then direct the build.
The one-person platform is no longer a fantasy. The question is whether you have the clarity to build the right one.
Lead the Build is a course for business experts who want to create internal tools and custom platforms using AI — no code required. Join the waitlist.