The question
Could a small business run largely through AI agents earn a first payment from someone outside my network? The first milestone is deliberately small: a real person choosing to pay for something useful.
The proposed experiment
The research brief starts with a hypothetical $100 budget. It asks an agent to explore opportunities, examine existing demand and alternatives, eliminate weak ideas, and model the strongest candidates. This page describes that design; it does not report sales or a completed launch.
What I want to measure
- Evidence that people already want the product or service
- Time and cost to the first genuine transaction
- Human minutes required to keep the business running
- Actual fees, fulfillment costs, refunds, and ongoing expenses
- The assumptions most likely to change the outcome
Where simulation helps
Bear, base, and bull scenarios can show how an idea depends on price, traffic, conversion, and costs. The inputs must be explicit. A model can help choose a test; it cannot stand in for customers or transactions.
The next evidence
The useful next result is a narrow candidate with an inexpensive real-world test and a pass/fail criterion set before launch. Future notes will distinguish proposed actions from completed tests and measured results.