LLMs make the "good enough" viable zone larger.

A product has to help users get to a "good enough" result quickly for it to be viable.

Before the user gets to a good enough result they are liable to give up at any moment and never return.

Many of the most powerful tools have a steep learning curve before they can output anything good enough.

That means they're only viable with a small, hyper-motivated market.

Once you have a viable product (it has achieved PMF with some small audience) you can move to hill climbing.

How much you have to plan and test before shipping is tied to how easy the "good enough" target is to hit for users, in a way that stands out from others.

For traditional software in a crowded area the target is very small.

LLMs allow much more forgiving tools that can get you a good enough result quickly.