Who it’s for
Four friends deciding which startup idea to pursue together.
Shops, friends and play · live
For four friends deciding which startup idea to pursue together. They argued over ideas without shared criteria. So each person rates every idea on the same four things, on their own, and a sealed second opinion waits until the humans have had their say.
Four friends share one room and one final scoreboard.
The other three have drafted theirs. Edit yours, or keep the suggestion.
Score the business, not the speaker. Each friend submits one rating per idea. Keys 1–5 set the focused row.
Presentation order is randomised, so nobody picks the slot after the strongest pitch.
A sample council has already read all four drafts. It stays sealed until every idea has been presented.
A simulation, not the app. The three friends, their ideas and their ratings are sample data, and the council review is a sample written for this page: in the real app five advisor calls, five anonymous peer reviews and a chairman run on my laptop. Nothing leaves your browser.
Four parts, plainly.
Four friends deciding which startup idea to pursue together.
They argued over ideas without shared criteria.
A shared session where friends draft ideas, rate each one on four criteria on their own, and present them in a random order. The human scoreboard is then compared with a sealed AI council review that stays hidden until someone reveals it.
Live, for the friends it was built for; public repository (named Startup_war).
The small print.
The council follows the LLM Council method credited in its README.
Honest line: the four criteria, the stage names, the random order and the council’s structure are the app’s own. The friends, their ideas, their ratings and the council’s words are samples.