Idk how far we go into regressions, afaik we don't really have an issue with it... We have a "everybody owns the product" mentality, and so pretty much everyone on the team should know how to use it and where issues might arise or have arisen.

So each code change needs to be kinda demonstrated that it works, whether it's through new tests, or screenshots, video, etc.

Usually if there's a specific bug that's been fixed, there is a test to catch if that bug occurs again. Most of the time it's been good enough.