A Two-Person Startup Made $1.8 Billion Selling Weight Loss Drugs — Using AI to Generate Fake Doctor Profiles and Before/After Images

Medvi was, for a brief period, celebrated as a triumph of AI-driven business efficiency. A two-person telehealth startup generating $1.8 billion in revenue — that number attracted the kind of coverage that frames a company as a proof point for what AI makes possible. The New York Times profiled it. The coverage emphasized scale and operational leverage. Then the methodology became public.

What Medvi Actually Built

Medvi markets GLP-1 weight loss medications — semaglutide and tirzepatide prescriptions, the class of drugs that includes Ozempic and Mounjaro — through a telehealth model. The business model itself is legitimate; the category has attracted billions in venture funding as obesity treatment demand has expanded. What distinguished Medvi was its marketing operation.

According to reporting on the company's practices, Medvi used AI to generate fabricated doctor profiles on social media platforms — synthetic physician identities with AI-generated headshots, credentials, and posting histories, deployed as organic-seeming endorsements of the company's products. The operation also produced fake before-and-after comparison images and AI-generated video content, all deployed at scale across digital marketing channels. The advertising strategy was not a supplement to legitimate marketing — it was apparently the primary growth mechanism.

The Efficiency Problem

The Medvi case exposes a genuine tension in AI adoption narratives. When a company achieves extraordinary efficiency metrics through AI, the first assumption is that it has found a better operational approach. The Medvi numbers — $1.8B in revenue, two employees — were extraordinary enough that the Times treated them as evidence of AI's transformative potential for lean businesses. The number is real; the methodology behind it is what the coverage missed.

The question the case forces is whether Medvi's revenue-per-employee figure was achievable through legitimate AI-powered marketing at similar scale, or whether the number is specifically a function of abandoning conventional advertising standards. There is a meaningful difference between AI enabling a two-person team to market effectively at scale, and AI enabling a two-person team to fabricate medical credibility at scale. The former is a business innovation; the latter is fraud.

Regulatory and Legal Exposure

The Federal Trade Commission's guidelines on AI-generated endorsements and testimonials are explicit: synthetic content presented as authentic customer or professional endorsements constitutes deceptive advertising. Fabricated doctor profiles that appear organic are the clearest possible violation of this standard. The healthcare context adds additional exposure — the FDA regulates pharmaceutical marketing, and synthetic physician endorsements for prescription medications create a direct regulatory liability.

The case is being cited as a cautionary example within the AI industry not because AI-powered advertising is inherently problematic, but because it illustrates how the efficiency and scale capabilities of generative AI lower the operational cost of fraud to a level that makes it economically attractive. The marginal cost of generating a hundred fake doctor profiles is near zero. The marginal cost of regulatory enforcement is not. Closing that gap is now an active concern for the FTC, FDA, and several state attorneys general who have cited the Medvi case in recent enforcement communications.

The Broader Signal

AI critics have argued since the beginning of the generative AI cycle that the technology's most significant near-term risk is not catastrophic misalignment but mundane misuse — fraud, disinformation, and deception at scale, enabled by systems that make synthetic content cheap and convincing. The Medvi case is that argument made concrete, in a domain — healthcare advertising — where the consequences of deception extend beyond financial harm to patients making treatment decisions based on fabricated medical credibility.