The startup org chart used to look like a wedding seating plan: founders, engineers, sales, marketing, finance, legal, HR, three advisors no one could identify and a “growth person” whose job was mostly interpretive dance with dashboards.
Now it looks more like a group chat: founder, co-founder, Claude, Stripe, Midjourney, one very nervous lawyer and a customer support bot that may or may not have invented a discount code while you were asleep.
Microcompanies are not new. Silicon Valley has always loved the tiny team with an absurd outcome. Facebook bought Instagram for $1 billion when it was hiring roughly 10 people, and WhatsApp sold to Facebook for $19 billion with 55 founders and employees serving 450 million monthly users.
What is new is the machinery. The old microcompany used cloud hosting and mobile distribution to avoid hiring an army. The artificial intelligence microcompany uses generative AI as the army: junior engineer, art department, ad agency, analyst, intern, call center and occasionally the overconfident guy in the meeting who should have stopped talking two slides ago.
The surge is real, even after discounting the LinkedIn confetti cannon.
The share of new startups with a solo founder rose from 23.7% in 2019 to 36.3% in the first half of 2025, according to Carta, which explicitly pointed to AI expanding what one person can build and sell.
A working paper from Harvard Business School and INSEAD added the structural proof. AI-tagged startups have about 25% fewer employees than comparable non-AI firms, with more engineers, fewer managers and similar valuations.
Translation: The modern startup is not just skipping ping-pong tables; it is skipping departments.
The cleanest answer to “How new is this?” is that the ambition is old, the operating system is new. OpenAI CEO Sam Altman’s now-famous bet about the first one-person billion-dollar company sounded like sci-fi until companies began treating AI less like software and more like staff.
The PYMNTS example is Medvi, a GLP-1 telehealth startup that Matthew Gallagher launched from his Los Angeles home with $20,000, no employees and a dozen-plus AI tools. Medvi reportedly posted $401 million in first-year sales, reached 250,000 customers, delivered a 16.2% net profit margin and is tracking toward $1.8 billion in 2026 revenue, with Gallagher’s brother as the only hire.
The caveat is important because the Weekender is cheerful, not gullible. Medvi outsourced licensed physicians, prescriptions, fulfillment, shipping and compliance; AI handled code, copy, ads, service and monitoring. Its chatbot also reportedly hallucinated prices and product lines, which is a reminder that a one-person company can become a one-person fire department very quickly.
The current scoreboard is dazzling, although “ROI” here really means reported revenue per employee, not profit after compute bills, contractors and platform tolls.
Medvi
Roughly $401 million in first-year sales with a two-person headcount means about $200 million per human, before counting outsourced clinical and logistics partners. That is less lean startup than startup wearing a jetpack and holding a clipboard.
Cursor/Anysphere
The AI coding assistant said in November that it crossed $1 billion in annualized revenue with more than 300 employees after raising at a $29.3 billion post-money valuation. That is more than $3 million in annualized revenue per employee. Bloomberg later reported that Cursor surpassed $2 billion in annualized revenue in February. In June, SpaceX announced it will acquire Cursor for $60 billion.
Lovable
The Swedish vibe coding platform hit $100 million in annual recurring revenue in eight months, then, according to TechCrunch, reached $400 million ARR with 146 full-time employees, or $2.77 million ARR per employee. By June, TechCrunch reported that Lovable surpassed $500 million in annualized revenue and was seeing 1 million new projects a week.
Midjourney
The viral math often gives Midjourney almost mythological productivity, but the conservative reported version is wild enough. The Information reported in 2023 that Midjourney was on pace for $200 million in annual revenue with 40 employees and no outside funding, while Forbes/PitchBook later put 2024 revenue at $300 million and said the company was profitable.
So yes, AI has enabled a new kind of microcompany. But the “company of one” is usually a company of one plus APIs, contractors, regulated partners, cloud vendors, model providers, payments rails and a stack of bots that never ask for PTO. The future may not belong to the founder who hires nobody. It may belong to the founder who knows exactly which humans not to hire yet.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Facts Only
* Startup organization historically involved founders, engineers, sales, marketing, finance, legal, HR, and three advisors.
* The contemporary structure resembles a group chat involving founders, AI models (Claude, Midjourney), software platforms (Stripe), and support bots.
* Facebook acquired Instagram for $1 billion when it had approximately 10 employees.
* WhatsApp was sold to Facebook for $19 billion with 55 founders and employees serving 450 million monthly users.
* The surge in solo founder startups increased from 23.7% in 2019 to 36.3% in the first half of 2025, according to Carta.
* AI-tagged startups have approximately 25% fewer employees than comparable non-AI firms with similar valuations, according to a working paper from Harvard Business School and INSEAD.
* Medvi achieved $401 million in first-year sales with a two-person headcount and reached a 16.2% net profit margin.
* Cursor/Anysphere achieved over $1 billion in annualized revenue with more than 300 employees after raising a $29.3 billion valuation.
* Lovable reached $400 million ARR with 146 full-time employees, equating to $2.77 million ARR per employee by June.
Executive Summary
The structure of a startup has shifted from a traditional organizational hierarchy to a decentralized, AI-augmented system. The operational model has evolved from a formal structure involving founders, specialized roles (engineers, sales, marketing, etc.), and advisors to an agile group environment heavily reliant on generative AI tools for execution. This shift is evidenced by the rise of microcompanies, demonstrated by historical examples like Facebook and WhatsApp acquiring smaller entities. The key difference now appears to be the machinery: instead of hiring large teams, modern startups utilize generative AI to simulate the functions of entire departments, including engineering, marketing, and support.
Evidence suggests that this trend is supported by data showing an increase in solo founders, with the share rising from 23.7% in 2019 to 36.3% in the first half of 2025, attributed partly to AI's ability to expand individual capabilities. Furthermore, studies indicate that AI-tagged startups have fewer employees than comparable non-AI firms while maintaining similar valuations. Concrete examples demonstrate this shift with companies like Medvi, which achieved significant sales and margins by outsourcing non-AI functions while using AI for core development and operations. The resulting economic measure often shifts the focus from traditional profit metrics to revenue per employee when factoring in outsourced services and platform costs.
Full Take
The narrative moves from an organizational story to a technological paradigm shift regarding production capability. The core implication is that the historical definition of a successful microcompany—defined by small physical teams—is being superseded by a definition based on leverage applied to digital, automated resources. The shift described is not merely about skipping traditional departments; it is about decoupling output from headcount through algorithmic orchestration. This introduces a critical tension between observed dazzling metrics (high revenue per employee) and the qualitative risk highlighted by the example of Medvi's dependency on outsourced services and chatbot hallucinations.
The concept that "the operating system is new" suggests that AI functions as a systemic layer, allowing for radical scaling without proportional human overhead in specific functional areas. The pattern observed is a tendency to reframe organizational inefficiency as a solvable engineering problem, which aligns with the historical Silicon Valley drive toward extreme efficiency. However, the challenge lies in distinguishing between genuine structural innovation and superficial automation—whether the delegation of tasks truly reduces risk or simply externalizes complexity onto less traceable partners. The focus must shift from the quantity of deployed AI tools to understanding who controls the emergent logic of these systems and what new forms of agency are being created in this post-departmental structure.
Bridge Questions: If organizational roles become largely obsolete, what new skills will define high-leverage human contribution? How does the reliance on opaque systems (like hallucinating chatbots) affect accountability when scaling rapidly? What mechanisms need to be developed to ensure that AI-enabled velocity translates into sustainable, equitable growth rather than ephemeral scale?
Sentinel — Human
The text is a sophisticated analytical piece that uses specific examples to build an argument about the structural shift in microcompany formation enabled by AI, exhibiting strong human-like synthesis rather than purely mechanical generation.