Some experience from dealing with this recently: Unless you have a very large design team and long timelines, it's best to start with an existing design system and tweak it. Adjust it to your liking, get buy-in, and then agree to keep it locked in place except for minor adjustments. Save the big goals for a V2 design to be launched at a later date.

A design system is only as valuable as the time it saves. If the design team spends months perfecting a design system and associated sandbox demo before they can even get to the core design work, it's unlikely that it's actually helping deliver the product on time.

It's also dangerous to let the design system become a moving target, where the design language changes from week to week. Each change will burn developer time integrating the changes, which will inevitably turn into multiple sprints dedicated to creating a theming system for your products, none of which really moves the product forward.

In a true startup environment, if you can't get the design system 90% complete in the first week or two, it's at risk of becoming more of a liability than a benefit.