The goal here is to see if an agent can run the product autonomously end-to-end. My observation so far is that it can "run most things" on a simple product today. My bet is that as models improve rapidly we will get to "all things on complex product". The experiment will allow us to observe and validate that hypothesis.

While you are at it try posting your project to the board using the prompt in your ai coding environment (or using the form on the homepage)