A Solo Founder Built a $401M Company with AI Tools and $20K. The One-Person Unicorn Is Real.

Matthew Gallagher launched Medvi, a GLP-1 telehealth startup, from his Los Angeles home in September 2024. Starting capital: $20,000. Employees: zero. Tools: a dozen AI products. Valuation after one year: $401 million, with margins surpassing incumbents in the space.

This is the proof point that connects every other story today. Agentic engineering (story 1) gives you the methodology. Quality tooling like Agent Skills (story 2) gives you the guardrails. Managed infrastructure at $0.08/hr (story 3) gives you the deployment layer. Affordable open models (story 4) give you the cost structure. Put them together and one person with $20K can build a company worth $401M.

Solo-founded startups now represent 36.3% of all new ventures according to Scalable.news, up dramatically from historical norms. At Anthropic's Code with Claude conference, Dario Amodei predicted the first billion-dollar single-employee company would arrive in 2026 with 70-80% confidence. Gallagher's $401M isn't a billion, but it's close enough to make the prediction feel conservative.

I shipped three solo products last year. Rayni, Document Domain Agents, Goldlink. 360K+ lines of production code, just me. None of them are worth $401M, but the workflow is the same. AI handles the volume. Human taste handles the direction. The bottleneck isn't writing code. It's knowing what to build and having the judgment to evaluate whether the output is good enough.

There's a counterpoint worth acknowledging. GLP-1 telehealth is a specific market with specific dynamics. Gallagher didn't build a general SaaS product. He built a telehealth company in a market with massive tailwinds. The AI tools reduced the team size from "need 30 people" to "need 1 person," but the market opportunity had to exist first.

Still. $20K to $401M in a year. Solo. That's not a rounding error. For builders: the constraint on what you can build alone has shifted. The question isn't "can I build this?" anymore. It's "should I, and for whom?"