How Small Can a Company Get?
Editorial statement All three cases below are compiled from public reporting and regulatory filings. None of them is a client of this company, and nothing here endorses their business model, financial position or compliance status. Every figure is sourced; numbers that are forecasts, unaudited or not peer-reviewed are labelled where they appear. References are listed at the end and link to the originating institutions ([1]–[10]).
Introduction: Three numbers, three truths
In 2026 the debate about the "one-person company" was driven by three numbers: $80 million, $401 million, and eleven labs.
The first came from a software company founded six months earlier and sold to Wix for $80 million in cash [1]. The second came from a telehealth company with two full-time employees and $401 million in annual revenue [7]. The third came from a biotech with one full-time employee coordinating eleven external laboratories [8].
All three numbers are seductive. Placed side by side, they show something else: none of these companies was "one person from beginning to end." One had eight employees; one was two brothers; one outsourced its entire R&D function to eleven labs.
What is worth studying is not "how big one person can build," but: what did each of them hand over, and what did each keep?
Case 1 | Base44: six months, eight people, $80 million
Background
Base44 is a tool that generates a complete application from a text description — part of the "vibe coding" category that emerged in 2025. The user types a description; the platform generates the database, storage, authentication, analytics and third-party integrations [1]. Founder Maor Shlomo was 31 at the time and had previously founded the B2B data platform Explorium.
Timeline
| Date | Event | Source |
|---|---|---|
| Late 2024 | Development begins as a side project | [1] |
| First three weeks after launch | 10,000 users | [1] |
| May 2025 | $189,000 profit for the month (after high LLM token costs) | [1] |
| 18 June 2025 | Wix announces acquisition for $80 million cash | [1][2] |
| At closing | 8 employees; $25 million of the price paid as a "retention" bonus, included within the $80 million | [1] |
Key numbers
- Users: 250,000 six months after launch; 10,000 in the first three weeks [1]
- Funding: zero outside capital (bootstrapped) [1]
- Profit: $189,000 in May 2025 — profit, not revenue [1]
- Price: $80 million cash, confirmed by Wix to TechCrunch [1][2]
- Staff: 8 people, confirmed by Wix [1]
The turn: it was not a one-person company
TechCrunch put it plainly:
"Admittedly, this wasn't a billion dollars or close to it. And Shlomo wasn't truly solo — he had eight employees, Wix confirmed. They will collectively receive $25 million of the $80 million as a 'retention' bonus." [1]
Note that many Chinese-language retellings describe the $25 million retention bonus as an addition on top of the price. That is wrong: it is part of the $80 million [1].
What it means for one-person companies
- Six months can produce a real business: users, profit, a buyer. This is not a "build for two years first" story.
- "One-person company" describes a starting point, not a promised end state: eight employees are not a failure, they are the cost of scaling.
- Keep the categories apart: profit and revenue differ by an order of magnitude; the bonus is part of the price, not a premium on top of it.
Case 2 | Medvi: two people, $401 million, and a warning letter
Background
Medvi is an online telehealth company that opened in September 2024, selling online consultations and mailed weight-loss medication (GLP-1 drugs and compounded semaglutide and tirzepatide) into a market with far larger incumbents. The New York Times published a profile on 2 April 2026 [5]; Forbes followed the same day [6]; Business Insider subsequently uncovered problems on the advertising side [7].
Key numbers (FY2025)
| Item | Figure | Source |
|---|---|---|
| Revenue | $401 million | [7] (citing NYT) |
| Profit | $65 million (about 16.2%) | [7] |
| Customers | 250,000 | [5][6] |
| Full-time staff | 2 (the founder and his brother) | [5][7] |
| 2026 forecast | $1.8 billion (company/founder claim, not audited) | [5][7] |
| First-month spend | $20,000 (marketing plus AI software) | [6] |
| Ad source | about 30% from affiliates (founder's statement) | [7] |
⚠️ Citation discipline: most Chinese reports describe the "$1.8 billion" as a valuation. It is not. It is a 2026 sales forecast, and the company has never raised equity [5][7].
Turn 1: the FDA warning letter (20 February 2026)
On 20 February 2026 the FDA issued a warning letter to MEDVi, LLC (dba MEDVi, Newark, Delaware), reference MARCS-CMS 721455, from the Center for Drug Evaluation and Research (CDER). It states that the website medvi.io, reviewed in December 2025, made "false or misleading" claims about its compounded semaglutide and tirzepatide products, in violation of sections 502(a) and 502(bb) of the Federal Food, Drug, and Cosmetic Act [4].
The letter quotes the site directly:
"Same active ingredient as Wegovy® and Ozempic®." "Same active ingredient as Mounjaro® and Zepbound®." [4]
The FDA's comment: compounded drugs are not FDA-approved, and these statements suggest the products are approved or have been evaluated for safety and efficacy when they have not [4].
Founder Matthew Gallagher told Business Insider that the website was operated by an affiliate he declined to name, who used his company name without authorisation, and that the affiliate had taken the site down and responded to the FDA [7].
Turn 2: AI-generated "doctors"
After reviewing Meta's ad library, Business Insider found at least six pages presenting themselves as doctors and promoting Medvi's weight-loss and men's products [7]:
| Page | Problem |
|---|---|
| Dr. Matthew Anderson MD | Listed an Angolan phone number; the page appeared to have belonged to a gospel singer |
| Dr. Spencer Langford MD | Older posts and contact details matched a clothing shop in the Republic of the Congo |
| Wade Frazer MD | Dropped "MD" after BI's enquiry; the same profile photo was used by three other pages |
| Several others | Profile photos carried a Gemini watermark, or the context was incoherent (e.g. a real-estate agent selling weight-loss drugs) — signs of AI generation |
At the time of BI's review, none of these pages carried a visible AI disclosure [7].
Ad volume: on the Friday there were still more than 5,000 active ads mentioning or linking to Medvi; after BI raised the issue with Gallagher, the count fell to about 2,800 by Monday [7].
Gallagher's response:
"In line with the FTC, we have a clear policy of providing disclosure on any actor or AI portrayal of a doctor or not using them at all. If we find an affiliate doing this we work to take these ads down." [7]
Turn 3: regulatory and litigation pressure
- In September 2025 the National Consumers League (NCL) and others petitioned the FTC to investigate; Medvi was one of six telehealth companies named [7]
- Nancy Glick, NCL's director of food, nutrition and obesity affairs, said Medvi's use of phrases such as "trusted by experts" and "doctor-approved" led consumers to believe its compounded drugs had been tested for safety [7]
- At least three lawsuits allege violations of spam laws together with affiliates (one withdrawn, two pending); the company denies wrongdoing [7]
What it means for one-person companies
- AI multiplies scale tenfold and reputational risk tenfold. Faking things got cheaper; being caught did not.
- Affiliate marketing is outsourced growth and outsourced risk: partners you do not control speak in your name, and regulators only recognise your name.
- Quote numbers with their qualifiers attached: $1.8 billion is a forecast, $401 million is a reported figure, and the regulatory status is a warning letter, not a ruling. Mixing the three is using a contested story to endorse your own product.
Case 3 | Mithril Medicine: one person, eleven labs, one drug pipeline
Background
Mithril Medicine, based in Cambridge, Massachusetts, is a "one-person company" acquired by Anthropic, focused on two rare muscular dystrophies: LAMA2-related congenital muscular dystrophy and alpha-dystroglycanopathy [8][9]. Anthropic disclosed the acquisition on 27 October 2026; terms were not disclosed [8][10].
Key facts
| Item | Detail | Source |
|---|---|---|
| Company size | One full-time employee (founder Cody Tranbarger), plus part-time scientific advisers | [9] |
| External labs | 11 | [8][9] |
| Claude's role | Coordinating contracts, logistics, experiments and much of the rest of operations | [8] |
| R&D speed | From idea to positive results in mice in under 18 months | [8] |
| Clinical timeline | LAMA2 programme first in-human trials as early as end-2028 | [8][9] |
| Second programme | alpha-dystroglycanopathy, as early as 2029 | [9] |
| Candidate selection | targeted for 2027 | [9] |
| Technical report | Not peer-reviewed | [10] |
What Claude actually did
Anthropic's own wording:
"Its founder, Cody Tranbarger, used Claude to organize contracts, logistics, and experiments with 11 external labs, and delegated many other aspects of Mithril's operations to Claude. As a result, Mithril took a possible treatment for one form of muscular dystrophy from an idea to promising results in mice in less than 18 months." [8]
The one disclosed efficacy data point: in the technical report, mice treated with a particular antibody design showed grip strength roughly twice that of untreated mice (the report has not been peer-reviewed) [10].
This person is not an ordinary person
Tranbarger graduated in 2025 from Harvard's MS/MBA biotechnology programme and previously worked at Goldman Sachs investment banking, the New York life-sciences investor Casdin Capital, and the Cambridge gene-editing company Prime Medicine [9]. In his public posts he calls this model "Pipeline-in-a-Person" [9].
Limits that must be stated
- Mouse results are not human results, and the technical report is not peer-reviewed [10].
- Anthropic itself says the work "remains early, with most of it preclinical," and most of the drugs have not entered human testing [8].
- This is an "AI amplifier" case: AI did not replace expert judgement, it let one person carry the pace of what used to be an entire R&D department — on the premise that he already knew what a good experimental design looks like and which data can be trusted.
What it means for one-person companies
- AI amplifies a person's expertise, not a person's ignorance: the same tool given to an expert and to a novice produces results an order of magnitude apart.
- "A department's work" can compress into "one person's scheduling", provided the process can be decomposed and external suppliers can be coordinated.
- High-barrier industries suit one-person companies: the barrier keeps competitors out, and AI offsets the scale disadvantage.
Analysis: three walls, one motion
| Case | What it broke | What it hit | The core motion it kept |
|---|---|---|---|
| Base44 | The time barrier in product development | Scaling still needs a team (8 people) | Compress the build; let word of mouth and community drive growth |
| Medvi | The speed barrier in growth | The compliance and trust floor (advertising ran out of control) | Outsourced growth to affiliates without building oversight in |
| Mithril | Both the expertise and the scale barrier | Expert judgement cannot be outsourced; results still unverified | Turned 11 labs into dispatchable units |
The one thing all three share: they are all doing "dispatch" — breaking work into assignable units, then deciding which unit goes to whom, when, and to what standard.
The difference is what gets dispatched. Base44 dispatched its own time. Medvi dispatched other people's traffic (and lost control). Mithril dispatched the capacity of eleven labs (and kept control, because an expert set the standards).
That line is the boundary between a "one-person company" and one person's chaos.
References
All accessed 6 November 2026 (HKT). Links point to the original publishing institutions; some outlets use paywalls or bot protection and may not open in every environment.
[1] Julie Bort, "6-month-old, solo-owned vibe coder Base44 sells to Wix for $80M cash", TechCrunch, 18 June 2025. https://techcrunch.com/2025/06/18/6-month-old-solo-owned-vibe-coder-base44-sells-to-wix-for-80m-cash/
[2] "Wix Further Expands into Vibe Coding with Acquisition of Base44, a Hyper-Growth Startup that Simplifies Web and App Creation with AI" (company release), GlobeNewswire, 18 June 2025. https://globenewswire.com/news-release/2025/06/18/3101508/0/en/wix-further-expands-into-vibe-coding-with-acquisition-of-base44-a-hyper-growth-startup-that-simplifies-web-and-app-creation-with-ai.html
[3] "Wix acquires Israeli vibe coding co Base44", Globes (English edition), 18 June 2025. https://en.globes.co.il/en/article-wix-acquires-israeli-vibe-coding-co-base44-1001513267
[4] U.S. Food and Drug Administration, "MEDVi, LLC dba MEDVi — MARCS-CMS 721455 — 02/20/2026" (warning letter), Center for Drug Evaluation and Research, 20 February 2026. https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/medvi-llc-dba-medvi-721455-02202026
[5] "How A.I. Helped One Man (and His Brother) Build a $1.8 Billion Telehealth Company", The New York Times, 2 April 2026. (Paywall) https://www.nytimes.com/2026/04/02/technology/ai-billion-dollar-company-medvi.html
[6] Josipa Majic, "How Medvi Found Success With Just $20,000 And AI", Forbes, 2 April 2026. https://www.forbes.com/sites/josipamajic/2026/04/02/ai-and-20000-helped-one-man-build-a-18-billion-telehealth-startup/
[7] "Medvi, the AI-powered telehealth company, is fueled by ads from doctors who don't appear to exist", Business Insider, April 2026. https://www.businessinsider.com/medvi-ai-weight-loss-millions-ai-advertising-legal-compliance-challenges-2026-4
[8] Anthropic, "Early progress on accelerating drug development with Claude" (official research post), 27 October 2026. https://www.anthropic.com/research/early-progress-drug-development
[9] "Anthropic buys Mithril Medicine, a one-person biotech", R&D World, 27 October 2026. https://www.rdworldonline.com/anthropic-acquires-mithril-medicine-one-person-biotech
[10] "Anthropic buys a one-person biotech and starts its own drug programmes", The Next Web, October 2026. https://thenextweb.com/news/anthropic-mithril-acquisition-claude-drug-development
Data notes (for editors and readers)
- Unverifiable figures are not used. Claims circulating online — such as "Claude organised 341 pages and 113 antibody constructs in 8 minutes" — are not included here: no matching text could be found in Anthropic's own article or in verifiable reporting.
- "$1.8 billion" is cited only as a forecast, and always alongside the fact that the company has raised no equity [5][7].
- Each monetary figure is labelled: the acquisition price is cash, the bonus is inside it; profit is not revenue; revenue is not valuation.
- Regulatory status is stated precisely: an FDA warning letter is not a ruling; pending litigation is pending, with no presumption of guilt.
- Non-peer-reviewed findings are flagged at least once and never described as "proven effective."