Domain Venture Factory
A research system for the business ideas behind a portfolio of domains, built so that the reading, the demand check and the review were done by separate agents with a person accountable for the call.
The problem it addressed
A portfolio of domains is a portfolio of unanswered questions. Each one needs research, a demand check, and an honest read on whether anyone is actually underserved. Done by hand that work is slow, and slowness is what lets weak ideas survive longer than they should.
So I built a tool for it — a terminal-based system that ran several agents across one domain at a time and handed back the evidence it had gathered, rather than a verdict.
How decisions were reviewed
Research, build and review were separate roles. One agent doing all three ends up marking its own work.
Every run got an isolated workspace, so a bad run could not quietly contaminate a good one.
A person was accountable for every decision. The system produced evidence and a recommendation; it did not get to approve itself.
Status
A past project, not currently active. It sits here because the separation of research from review is the part worth carrying into other work.