Can one person run this company
- Learning goals
- List the jobs a founding team used to split. Decide for each whether you do it, an agent does it, you buy it in, or it waits.
- Expected outcomes
- A completed Job Map and a first view of which gaps the later units must close.
Why founders worked in teams
Start-ups have usually been founded by two or three people because the work split naturally between them. One common pairing was a business founder and a technical founder; a larger team might divide product, engineering and go-to-market. Common investor wisdom has treated a solo founder as a warning sign.
That split existed because one person could not build the product, sell it and run the company at the same time. AI tools change part of that calculation.
What AI changes
With coding assistants and agents, one person can now ship a first version of a product that once needed a technical co-founder. People who could never code, from finance, consulting or the creative industries, can reach a working prototype on their own. Domain experts, such as a doctor, a property broker or a teacher, can build software around the problems they see every day.
What AI leaves in place
The case against going solo is strong, and this school is built around it.
- Teams get the same boost. If AI multiplies every builder's output by the same factor, a team of three still does roughly three times what one person does.
- The bar rises. Once everyone can build more, investors expect a seed-stage product to be further along, so the solo founder gains no easier ride.
- Code was never the only constraint. Someone still has to hold the customer meetings. Sales, recruiting, judgment and persistence stay hard even as coding gets easier.
- A demo is not a product. Anyone can now produce a first version. The distance from a demo to a dependable product stays wide, and technical judgment about what to build counts for more.
Going solo works when you cover, deliberately, the jobs a co-founder used to cover.
This section summarises the arguments in Andrew Chen's essay New founder archetypes in the AI era. Read the essay for the full case for and against each founder profile.
Jobs in a company
The school uses nine jobs. Every company does all of them, whether or not anyone has named them.
| Job | What it covers | Unit |
|---|---|---|
| Build | Making and maintaining the product | 2 to 4 |
| Check | Catching errors before customers do | 4 |
| Sell | Talking to customers, pricing, closing | 5 |
| Reach | Getting found: content, channels, partners | 5 |
| Serve | Support and keeping customers | 2, 5 |
| Money | Costs, books, tax, paying yourself | 6 |
| Set up | Company structure, contracts, filings | 7 |
| Comply | Data protection, AI rules, security | 8 |
| Decide | Strategy, priorities, what not to do | Next edition |
Four ways to cover a job
- Me. You do it yourself. Keep jobs here when they rest on your judgment or your relationships.
- Agent. An AI agent does it and you check the result. Suits work that repeats and that you can verify.
- Buy in. A freelancer, accountant or lawyer does it. Suits work that needs a licence or carries legal liability.
- Later. The job can wait. Write down when it stops being able to wait.
Agent and Buy in still leave the job with you. If an agent sends a wrong invoice or an accountant misses a deadline, the company answers for it. Units 3 and 4 cover how to specify and check delegated work.
Exercise: your Job Map
For each job, choose who covers it now and add a note. Your answers stay in this browser and appear in the Planner.
Self-check
Further reading
- Andrew Chen, New founder archetypes in the AI era
- US Small Business Administration, Business Guide, plan your business
- Hugging Face, Agents Course, a preview of Unit 2