Incumbents
Built before AI. They use tools in an existing org and its headcount. AI makes people faster, but the structure and the cost base stay the same.
Autonomous business
An Autonomous Business (L5) runs everything in go-to-market, product development, and is self-improving.
Even at L5, human involvement doesn’t have to stop. The business can run on its own, while people can still work alongside it, shape its direction, and expand what it can achieve.
Autonomous, AI-native, and incumbent
Incumbent and AI-native describe how a company was built. Autonomous describes how it operates. Either kind of company can move toward autonomy.
The book uses ChatGPT’s public launch in 2022 as a practical dividing line for this comparison.
AI-native
Built with AI at the core from day one. Lean teams, AI in every workflow, but people still operate and direct the business day-to-day.
Autonomous
The end state. AI agents run and improve the business with little or no human operation. Growth finally decouples from headcount.
The Autonomous Business Scale
Five levels, from AI as a personal tool to a business that runs and improves itself. Developed through the research and conversations with founders and leaders behind Growth After Headcount.
L1AI-assistedPeople use AI to work faster. The org chart is unchanged.
Individuals reach for AI to lift their own productivity. A marketer brainstorming a post, an accountant checking a VAT rate, a support rep querying the knowledge base. The work is still theirs.
L2AI-draftedAI does the work; a person signs off before it ships.
AI produces work on a person's trigger for their validation. It runs the SEO research, drafts the invoice, suggests the reply, finally a human reviews, then sends.
L3AI-operatedAI runs whole processes. People handle the exceptions.
AI automates an entire process end to end. Publishing SEO articles, invoicing, first-line support. People move up to the complex cases and leave the simple ones to the system.
L4AI-runAI runs everything except the human moments.
AI executes all digital activity. Humans remain only for the highly personal. In-person events, the important client meeting, or the personal brand.
L5AutonomousThe business runs itself and improves itself.
AI runs the business from marketing to selling to invoicing, and repeats the cycle. No people are needed to run it. An autonomous business can improve itself.
The scale explained
While researching and interviewing for Growth After Headcount, it became clear that companies are on very different AI journeys. The Autonomous Business Scale gives those differences a shared language and helps a company identify its next step.
At L1, people use AI but still own the work. At L5, AI agents run the whole business and can improve it themselves. The owners can distance themselves from day-to-day operation.
Start with a complete workflow: who initiates it, who does the work, who approves the result, and who improves it? A business reaches a level when that describes how it operates across its functions, not just one automated task.
L3AscendAI-operated growth L4Swan AIAn automated go-to-market engine L5RB2BA business that runs itself
Examples discussed in the conversations with founders and leaders. Listen to the conversations for the context behind each placement.
Revenue per employee
Growth becomes more interesting when a company can generate more revenue without adding people at the same rate.
Annual revenue ÷ Average number of employees
What good looks like
A0 is the traditional, pre-AI baseline in this framework: below $500K in annual revenue per employee. People do most of the work, and growth generally means adding headcount. A1 starts at $500K per person, with each band setting a more ambitious revenue-per-employee target.
A0<$500K
A1$500K–$1M
A2$1M–$10M
A3$10M–$100M
A4$100M–$1B
A5$1B+
L1–L5 describe how a business operates. A0–A5 describe its revenue per employee. A0 is a reference point for traditional businesses; expectations vary by industry and business model. A high revenue figure alone does not make a company autonomous.
Compare businesses with similar models, and read revenue per employee alongside profitability, customer outcomes, contractor costs and software spend.
Who is already there
| Band | Company | Revenue / employee | Year |
|---|---|---|---|
| A0 | Target | ~$252K | 2025 |
| A1 | Super.com | $1M | 2025 |
| A2 | RB2B | $8.6M | 2026 |
| A3 | Midjourney | $12.5M | 2025 |
| A4 | Medvi | $401M | 2025 |
| A5 | The next frontier | $1B+ | — |
Target: $104.780B in fiscal 2025 net sales ÷ 415,000 year-end employees ≈ $252K per employee. Sources: 2025 financial statements and employee count. This estimate uses year-end headcount, including part-time and seasonal staff, rather than the annual average. A0 illustrates traditional business economics; it does not mean the company uses no AI.
A1–A4 source: Henry Shi’s Lean AI Leaderboard . Figures are snapshots; follow the full leaderboard for the latest data.
Where could your business go next?
Explore the conversations and decisions behind the framework.
Explore the podcast