The default answer is no

When a new request comes in, our first answer is no. Not to reject the idea, but to make sure it is truly necessary. We have seen too many features that looked great at first but became nothing but overhead over time.

Saying yes needs a reason. Before we say it, we ask: does work actually stop without this feature? Is it needed right now? Is there a simpler way to solve the same problem?

Most requests quietly disappear during this process. The ones that survive are the ones worth building.